WO2010130288A1

WO2010130288A1 - Method for operating a cylinder, system and mannequin robot comprising the same

Info

Publication number: WO2010130288A1
Application number: PCT/EP2009/055734
Authority: WO
Inventors: Richard Bernard Schaper
Original assignee: Richard Bernard Schaper
Priority date: 2009-05-12
Filing date: 2009-05-12
Publication date: 2010-11-18

Abstract

The present invention relates to a method for controlling piston displacement of a cylinder. It further relates to a system and to a mannequin robot comprising such a system. The method according to the invention is characterized in that a stable position of the piston (1) can be achieved by connecting both the first and second inlet (3,4) to the source of pressurized medium. The invention is applicable to both pneumatic and hydraulic systems.

Description

Method for operating a cylinder, system and mannequin robot comprising the same

The present invention relates to a method for controlling piston displacement of a cylinder. It further relates to a system and to a mannequin robot comprising such a system.

Cylinders, such as pneumatic or hydraulic cylinders, are pressurized medium piston-cylinders which move in and out, often referred to as the retracting and extending motion of the piston. The movement is achieved by injecting a pressurized medium, such as air, or gas in general, or hydraulic fluid such as water or oil, in one inlet of the cylinder while the medium is released through another inlet on the other side of the cylinder. This type of movement is between two extremes of the cylinder, e.g. between full retraction and extension. During this motion, the piston will cover the entire stroke range. The supply of pressurized medium can be interrupted during this motion to cause the piston to stop. However, this approach does not allow a precise control and or knowledge of the resulting position of the piston.

It is known in the art that stopping at an intermediate position of a pneumatic cylinder can be achieved by using switchable non-return valves, as indicated in figures IA and IB.

Figure IA depicts a pneumatic system comprising a double-acting pneumatic cylinder 1 with a piston 2. Cylinder 1 is provided with inlets 3, 4 for transport of pressurized gas. Each inlet 3, 4 is connected to a switchable non-return valve 5, 6. Both valves are connected to switchable valve 7. Valve 7 can be used to change the connection of the valves 5, 6 to either an exhaust of pressurized gas available at terminals 8 and 9 or a source of pressurized gas available at terminal 10.

In figure IA, valve 7 is in a state in which both switchable non-return valves 5, 6 are connected to exhaust 8, 9. As a consequence, both switchable non-return valves 5, 6 are closed, preventing gas from leaving inlets 3, 4.

In figure IB, valve 7 is shifted to the right causing switchable valve 5 to be connected to the source of pressurized gas. Consequently, gas will flow through non- return valves 11, 12 towards inlet 3. Simultaneously, the pressurized gas will open non-return valve 13 allowing gas to flow from inlet 4 through non-return valve 13 and flow control valve 14 towards exhaust 9. Consequently, piston 2 will move towards the right. Piston 2 can be stopped by returning valve 7 to the state illustrated in figure IA. This can be used to stop in between the extremes of the stroke range of the piston or to have a safe stop in case of an emergency, if for instance the pressure or a controlling signal drops or when an emergency stop is pressed.

The disadvantage of this approach is that the position of the piston within the stroke range is unknown. Another disadvantage is the so-called 'jumping' of the pressure if the movement direction of the piston is changed. This is related to the difference in available surface of the piston for extending and retracting strokes and to the difference in pressure on both sides of the piston. This effect can give a shock or delay in moving the piston. A load on the cylinder will make this effect even worse. This effect will also appear when the piston reaches the end of the cylinder, e.g. one of the extremes of the stroke range. The pressure on the release side will drop to zero. When the pressurized gas is again applied to the piston there will be some delay because the area needs to be pressurized. This will result in a possible overshoot of the piston. It should be noted that hydraulic systems have similar disadvantages.

In general, the piston position can only be known accurately by employing servo feedback, the so-called servo pneumatics or hydraulics. However this approach is expensive and large and it needs a lot of hardware. Hence, servo pneumatics or hydraulics is a complex and costly solution if one desires accurate piston control. This is especially true in situations where a lot of cylinders are needed like in robotics applications, such as a mannequin robot.

It is therefore an object of the present invention to provide an alternative solution to servo pneumatics or hydraulics, which is less costly, requires fewer components, but still provides sufficient position control of the piston .

At least one of these objects is achieved with the method for controlling piston displacement of a cylinder according to the invention. This cylinder has a first and second inlet corresponding to a retracting and extending stroke of the piston of the cylinder, respectively.

According to the invention, the method comprises the steps of providing a source of pressurized medium, providing an exhaust, and providing a valve unit for connecting the source of pressurized medium, the exhaust and the first and second inlet. It further comprises controlling the valve unit to switch to a retraction connection state for achieving the retracting stroke, in which the first inlet is connected to the source of pressurized medium and the second inlet is connected to the exhaust, or to an extension connection state for achieving the extending stroke, in which the second inlet is connected to the source of pressurized medium and the first inlet is connected to the exhaust. The method according to the invention is characterized by locking the piston by switching the valve unit to a lock connection state in which both the first and second inlet are connected to the source of pressurized medium.

Unlike the prior art approaches, locking of the piston is achieved by connecting the source of pressurized medium to both the first and second inlet. This prevents the difference in pressure that exists in the prior art when moving from the lock connection state to the extension or retraction connection state.

In prior art systems, the cylinder is actuated by switching from a state in which both inlets are connected to the exhaust, to a state in which only one inlet is connected to the exhaust whereas the other inlet is connected to a source of pressurized medium, e.g. gas. One could therefore say that in prior art systems, the cylinder is actuated by applying pressure. However, according to the present invention, the cylinder is actuated by switching from the lock connection state, in which both inlets are connected to the source of pressurized medium, to a state in which one inlet is connected to the source of pressurized medium whereas the other inlet is connected to the exhaust. One could therefore say that according to the invention, the cylinder is actuated by applying exhaust instead of pressure, whereas the system is pressurized in the lock connection state. This avoids the need to build up pressure prior to actuating the cylinder.

In case two separate valves are used, one for each inlet, tube length can be kept at a minimum. Hence, the amount of space outside the cylinder is low, which further improves the responsiveness and increases the stiffness of the system. The above mentioned effect can further be improved if, during the locking of said piston, medium is prevented from exiting the first and second inlet by using at least one non-return valve. The cylinder preferably comprises a pneumatic or hydraulic cylinder, in which case the source of pressurized medium comprises a source of pressurized gas or liquid respectively. Combinations of these systems are not excluded. The speed of the piston can be controlled by regulating the speed at which the pressurized medium exits the first or second inlet. This, together with the friction encountered by the piston itself, will determine the speed of the piston during displacement. Increasing the pressure on the other side of the piston will obviously again result in higher speeds of the piston.

It is advantageous to regulate the pressure of the pressurized medium that is supplied to the first and or second inlet. This makes it possible to compensate for the difference in effective piston surface that is available for the different stroke directions. Furthermore, as mentioned above, the speed of the piston can be regulated.

Preferably, it is detected whether the piston is within a predefined region in the stroke range of the cylinder. For instance, a reed switch can be attached to the cylinder. In this case, the piston is provided with a magnetic member. The reed switch will detect whether this member is in close vicinity. The reed switch has a certain length in which it will provide a signal indicative of the presence of the magnetic member. To which region of the cylinder the length corresponds depends on the placement of the reed switch on the cylinder. It is also preferable to determine whether the piston is at a predefined position in a stroke range of said cylinder. Unlike a region detector, like a reed switch, a position detector can also be used. For example, a medium, e.g. gas, sensor can be coupled to the inlet (s) . If the piston reaches one of the ends of the cylinder, i.e. an end of the stroke range, no more medium will leave the corresponding inlet. In this way, the medium sensor can be used to detect the extreme positions of the piston within the stroke range. It should be noted that a reed switch can also be used to determine whether the piston is at a corresponding position within the stroke range. If the piston moves into the detection region of the reed switch, the switch will change its output. This change is indicative that the piston has reached a position corresponding to an outer end of the detection region of the reed switch. As noted before, this position depends on the placement of the reed switch and can therefore vary during production or during operation due to adjustments by the user. However, the reed switch may also be intentionally moved. As will be explained later, it can be advantageous to divide the stroke range in a number of regions, in which a boundary between adjacent regions corresponds to an end of the detection region of the reed switch. By controlling the cylinder to move to a certain position relative the length of a region, for instance controlling the cylinder to move to 20% of region 1, the system is less dependent on the actual position of the reed switch. Furthermore, the position of the reed switch can be manually adjusted to change the motion characteristics without changing the control of the cylinder .

Preferably, controlling the piston displacement comprises applying the exhaust to the first or second inlet for a predetermined amount of activation time to displace the piston from a start position to a target position. In this case, the method comprises providing a calibration with calibration parameters of the cylinder, said calibration describing the motion characteristics of the piston during piston displacement. A virtual position is computed that corresponds to the start position of the piston. If this position is known and or measured, the virtual position can be set to this position. Otherwise, the virtual position may be calculated using the calibration and at least one previous piston displacement with respect to a known position of the piston.

Subsequently, a difference between the virtual position and the target position is calculated, which together with the calibration forms the basis for determining an activation time for the cylinder. The cylinder is then controlled using this determined activation time. During and or after the piston displacement, the virtual position is updated using the calibration and the amount of elapsed activation time.

In servo pneumatic or hydraulic systems there is a constant feedback of the position of the piston back to the valve control unit. This requires additional costly hardware. Instead of this control, which is driven by detected positions, the present invention is based on time control .

The exhaust is applied for a predetermined amount of time. Because there is generally not a direct measurement of the position of the piston, the concept of a virtual position is used. This virtual position is derived from a known position, e.g. at one of the ends of the cylinder or a position corresponding to the location of a position detector, in combination with previous piston displacements and a calibration of the cylinder. Given a known position, the influence of a piston displacement, or in other words the predetermined amount of activation time used to achieve this displacement, can be calculated once the behaviour of the piston is known.

According to the present invention, a calibration is used that comprises calibration parameters that describe different characteristics of the piston during piston displacement, e.g. speed or a derivative thereof. The calibration is therefore used to transform the predetermined amount of activation time in a spatial displacement of the piston .

After the exhaust has been applied to achieve the desired displacement, the virtual position should reflect the new position. This can be achieved by updating the virtual position during the displacement of the piston or updating the position only after the target position has been reached.

If a system comprises more than one system with different sizes it becomes cumbersome to keep track of all the spatial coordinates of possible positions. It is therefore more convenient to express the start and target positions relative to one of the stroke range and the region length. This eliminates the need for actually knowing the region length or the stroke range. For instance, a fully retracted piston could be indicated as "0", whereas a fully extended piston could be indicated as "1". Intermediate values, such as "0.5" would then indicate a percentage of the total stroke range (in this case 50 percent) . It is also possible to apply this approach to regions, e.g. "0" and "1" correspond to the start and end of a given region, respectively. Furthermore, the motion of the system can simply be adjusted by moving the position of the reed switch without having to change the settings of the control for the cylinder .

Once all positions are expressed in the above mentioned relative fashion, information is needed on the amount of activation time required to traverse a certain relative distance of the total stroke range, for instance from "0" to "0.2". To this end, it is advantageous if the calibration parameters comprise the amount of activation time needed for piston displacement between two known and or detectable positions. Similarly, the ratio of the relative distance and the corresponding required time could be referred to as the (relative) speed of the piston.

Other calibration parameters in the calibration could for instance be one or a combination of a start delay time, corresponding to a delay in time between controlling the valve unit and resulting motion of the piston, an acceleration time corresponding to the amount of time needed for acceleration from a standstill of the piston to a corresponding steady-state velocity, and a deceleration time corresponding to the amount of time needed for deceleration from a corresponding steady-state velocity of the piston to a standstill. These calibration parameters can all be used to account for non-ideal effects of the piston motion. Ideally, the piston should move at a constant (steady-state) speed for a predetermined amount of activation time, start and stop with infinite acceleration and deceleration, and respond instantaneously to commands for the valve control unit .

The calibration parameters may be determined individually or in combination. For instance, a motion of the cylinder will contain the acceleration delay and start delay. The effect of these times on the general behaviour may be described using a single parameter. Other calibration parameters can be extracted by using measurements of time needed by the piston to traverse between two known or detectable positions. The positions must be known or verifiable, in contrast to the virtual position, because the former positions are used to determine the virtual position itself.

As an example, time can be measured that is required to complete an extending and or a retracting stroke between two known or detectable positions in a single stroke. In addition, time can be measured that is needed to complete an extending and or a retracting stroke between the same two known or detectable positions but then in a sequence of strokes. These measurements can be compared and used for extraction of at least one calibration parameter or a combination thereof as previously mentioned. Once extracted, the extraction result, e.g. the acceleration and deceleration time, can be stored in the calibration.

An alternative or additional option to determine a calibration parameter is to measure the time needed to complete an extending and or a retracting stroke between two known or detectable positions in a single stroke, and to store this measured time as a calibration parameter in the calibration. Assuming ideal motion characteristics of the piston, this calibration parameter could suffice. This calibration parameter can be complemented by also storing the distance, and preferably also the direction, that was covered in the measured time. Additionally or alternatively, a speed calibration parameter is computed and stored by dividing the difference between positions and the corresponding required time, preferably relative to the total stroke range or region length. For instance, a measured time of 0.5 seconds to move from a relative position of 0.1 to 0.6 (extending stroke) would result in a speed calibration parameter of (0.6 - 0.1) /0.5s = 1/s. Usually, a cylinder is subjected to a sequence of piston displacements in between the extremes of the stroke range. According to the invention, the cylinder is subjected to a sequence of piston displacements, each having a start and target position. For instance, a cylinder could move from 0 to 0.2 to 0.4 to 1.0 (all relative positions) . In this case it is possible that one of these piston displacements corresponds to and or comprises a piston displacement needed for the construction or computation of the calibration. In that case, it is advantageous if the relevant calibration parameter is determined and stored. For instance, the displacement of the piston could comprise one of the above mentioned single or sequence of strokes. This approach does of course not exclude the possibility of performing the calibration beforehand. However, in certain applications, such as a mannequin robot, this feature of the present invention is particularly useful as it allows the system to be adaptive. During the execution of a motion program, which consists of a collection of sequential steps, the system is able to calibrate itself and is therefore capable to respond to environmental changes, such as a different loading of the cylinder, different humidity and temperature, etc. In case of a mannequin robot, one could therefore easily replace the clothing on the mannequin without the need to completely recalibrate the system.

It is advantages if the virtual position is set equal to the predefined position when the piston is at said predefined position. This predefined position could be associated with a position or region detector. For instance, during the motion of a piston from an intermediate position towards an end of the cylinder, the piston passes the position at which the position detector is located or the position which the detector is responsive to. At the moment the piston is at that position, the virtual position should ideally correspond to the true position, which in that case can be verified by the position detector. However, inaccuracies in the calibration and control of the cylinder may result in small errors. These errors can be eliminated by equating the virtual position and the true or predefined position. It is further possible to recalculate or adjust the predetermined amount of activation time to account for the change in virtual position.

Most cylinders, such as the double-acting pneumatic or hydraulic cylinders, are not symmetric in the sense that the motion characteristics of the piston are different for the retracting and extending stroke. It is therefore advantageous if the calibration comprises separate calibration parameters for the retracting and extending strokes thereby forming a bi-directional calibration. For instance, the aforementioned relative speed could be split into a relative speed for retracing strokes and a relative speed for extending strokes. In this case, it is preferable to perform the updating of the virtual position and the determining of the activation time in dependence of the motion direction of the piston, e.g. retracting or extending, using the corresponding bi-directional calibration .

Another or an additional advantage can be achieved if the stroke range of the cylinder is divided in at least two regions, each region being characterized by a corresponding bi-directional calibration. For instance, the total stroke range can be divided in two regions of equal size, such as an inner region and an outer region. If the updating of the virtual position and the determining of the activation time are done in dependence of the motion direction of the piston and the region in which the piston is positioned using a corresponding bi-directional calibration, one can accommodate for different loading depending on the position of the piston. If the piston is for instance used between the torso of a mannequin and one of its arms, the loading will differ with the position of the arm (piston) . In fully retracted state, and assuming a hinged connection of the cylinder to both torso and arm, the load will push the piston outwards, whereas in fully extracted state, e.g. the arm pointing in the air, the load may push the piston inwards. By having at least two regions with corresponding bi-directional calibrations, this problem can be obviated because the impact of the load can be compensated and or accounted for in the calibration.

Additionally, the boundary between regions can be used as a reference point. For instance, if a reed switch is used, one of the ends of the detection region can correspond to a reference point. An example could be the turning angle of a head of a mannequin robot. By displacing the reed switch, the reference point can be shifted. The reference point itself can serve as a point relative to which the motion of the cylinder is performed.

During displacement, the piston may traverse one or more different regions. It is therefore advantageous if the updating of the virtual position and the determining of the activation time is done in dependence of any region the piston may traverse during said piston displacement. For each region a corresponding bi-directional calibration may be used.

Usually, it is convenient to have a position and or region detector placed at the boundary between adjacent regions because gives a verifiable position at one end of the region, the other mostly being one end of the total stroke range or an end of an adjacent region.

The present invention also provides a system comprising a cylinder having a first inlet and a second inlet corresponding to a retracting and extending stroke of a piston of the cylinder, respectively. The system also comprises a source of pressurized medium, an exhaust, a valve unit for connecting the source of pressurized medium, the exhaust and the first and second inlet, and a valve controller for controlling the valve unit to switch between connection states. The system can be operable in a retraction connection state for achieving the retracting stroke in which the first inlet is connected to the source of pressurized medium and the second inlet is connected to the exhaust, and in an extension connection state for achieving the extending stroke in which the second inlet is connected to the source of pressurized medium and the first inlet is connected to the exhaust.

The system is characterized in that it is further operable in a lock connection state for locking the position of the piston in which both the first and second inlet are connected to the source of pressurized medium. As described above, this solution prevents inter alia the previously mentioned "jumping effect" when moving from the lock connection state to the extension or retraction connection state .

The cylinder preferably comprises a pneumatic or hydraulic cylinder, in which case the source of pressurized medium comprises a source of pressurized gas or liquid respectively. Combinations of these systems are not excluded. The system can further comprise at least one non-return valve in between the source of pressurized medium and the valve unit. This prevents any medium to flow out of the inlets back to the source of pressurized medium. This can for instance be useful if the source of pressurized medium is no longer available, for instance due to a failure, as a consequence of which the pressure sharply falls. Additionally, applying an external force to the cylinder will in this case not result in significant displacement of the cylinder. The non-return valves prevent or limit any uncontrolled motion of the piston.

A speed regulator can be placed in between the first and or second inlet and the valve unit. This allows speed control of the piston during the extracting and or extending stroke.

Additionally or alternatively, a controllable pressure regulator to regulate the pressure of the pressurized medium which is applied to the first and or second inlet may be used. This allows for compensation of any difference in effective piston surface. It could even compensate a constant loading of the cylinder, e.g. an external force exerting a pulling force on the piston. Assuming equal friction and speed regulation for both motion directions, identical piston speeds can be obtained. In a preferred embodiment of the present invention, the valve unit of the system comprises a first valve arranged to connect the first inlet to the source of pressurized medium in the retraction and lock connection state and to connect the first inlet to the exhaust in the extension connection state. It further comprises a second valve arranged to connect the second inlet to the source of pressurized medium in the extension and lock connection state and to connect the second inlet to the exhaust in the retraction connection state .

As mentioned previously, it is advantageous if the system comprises a region detector to detect whether the piston is present in a predefined region in a stroke range of the cylinder and or a position detector to detect whether the piston is at a predefined position in a stroke range of the cylinder.

The position detector could comprise a medium sensor connected to the first and or second inlet, which is arranged to measure the corresponding medium volume output. It is also possible that the piston is equipped with a magnetic member and that the region and or position detector comprises a reed switch coupled to the cylinder to detect said magnetic member. Both types of detectors have been described above.

In a preferred embodiment of the present invention, the valve controller comprises an input unit for receiving instructions to displace the piston from a start position to a target position by applying the exhaust to the first or second inlet for a predetermined amount of activation time. This input unit could be a digital interface with which the valve controller can be connected to a personal computer. However, it could relate to a different aspect of a programmed microcontroller. The valve controller further comprises a calibration memory for storing calibration parameters of a calibration, wherein the calibration describes the motion characteristics of the piston during piston displacement. Additionally, it comprises a virtual position unit arranged for computing a virtual position corresponding to the start position of the piston. This virtual position may be stored in a virtual position memory. The valve controller also contains an activation time calculate unit for calculating the required activation time based on a difference between the virtual position and the target position and the stored calibration. The virtual position unit is further arranged to update the stored virtual position using the stored calibration and the amount of elapsed activation time during the piston displacement. The system or the valve controller in particular, may comprise a calibration unit to determine at least one of the calibration parameters previously defined using the method as described previously. This calibration unit is preferably arranged to update the stored calibration if the displacement of the piston between the start and target position comprises the single stroke and or sequence of strokes used for the construction or computation of the calibration. This allows the system to be adaptive as described above.

The present invention may be used in a mannequin robot. These robots are normally used for displaying clothes. It is very attractive for stores to place these robots in store windows. People passing by will likely notice the moving robots thereby also drawing attention to the clothes that these robots are wearing. In addition, replacing static mannequins with robots according to the invention increases the vividness of the display. The cylinders are in this case connected between parts that should move relative to each other, such as between a head and torso, between the legs and torso, and between the lower arm and upper arm.

Because a lot of cylinders must be used to allow for smooth and life-like motion of the robot, the system must be low cost, robust and easy to calibrate and or install. The present invention provides such a system. Next, the present invention will be described in more detail under reference to the accompanying figures. In these figures, which relate to embodiments targeted at pneumatic systems but equally applicable to hydraulic systems, like numerals will be used to designate like or identical components, and wherein:

Figures IA and IB illustrate a prior art pneumatic system;

Figure 2 shows an embodiment of a pneumatic system according to the invention;

Figure 3 shows a time-velocity graph corresponding to piston movement;

Figure 4 shows a flow-chart corresponding to a movement from a start position to a target position; Figure 5 displays a flow-chart corresponding to updating the virtual position;

Figure 6 illustrates a pneumatic cylinder in which the total stroke range is divided in two regions; and

Figures 7A-7C demonstrate how to calculate calibration parameters using single and multiple stroke motion.

Figure 2 shows an embodiment of a pneumatic system according to the invention. Pneumatic cylinder 1 has two switching valves 15, 16 that control the movement of piston 2. In the default (locking) position, both switching valves 15, 16 connect the pressurized gas through non-returning valves 17, 18 that are placed in between a source of pressurized gas 19 and the switching valves 15, 16. By using non-return valves on the pressure supply, the pressurized gas, such as compressed or pressurized air, can flow in the pneumatic cylinder but cannot flow back. This will also happen after the cylinder stops at the end of stroke range. By keeping the cylinder under pressure it moves smooth and it provides a good stiffness. Thus, the piston performs a retracting stroke when valve 15 connects inlet 3 to the exhaust 20 and when valve 16 connects inlet 4 to the source of pressurized gas 19. The piston performs an extending stroke when valve 16 connects inlet 4 to the exhaust 20 and when valve 15 connects inlet 3 to the source of pressurized gas 19. Piston 2 is stopped when both valves 15, 16 connect the respective inlets 3, 4 to the source of pressurized gas 19.

Piston 2 generally has a different surface on each side. The surface on the shaft side is smaller. Due to this, the pressure on the larger surface needs to reduced to balance the force with that of the smaller surface. To this end, a pressure regulator 21 is present in the path between the source of pressurized gas 19 and inlet 3. Piston 2 can be brought to a quick and tight stop. It is therefore possible to estimate its position accurately with a model. When pressure is released from one of the chambers of the cylinder, piston 2 moves in the direction of that chamber at a fairly constant speed. Thus by knowing that speed and a reference/ initial position, the position of piston 2 can be calculated. This speed can also refer to the previously described relative speed, e.g. the speed relative to the stroke range or region length.

Generally, the pneumatic system does not have means to validate or verify every piston position by measurement.

Therefore, the system itself must estimate the position of the piston. According to the present invention, this can be achieved by using the concept of a virtual position. Ideally, this position should correspond to the true position of the piston.

Figure 3 shows an idealized time-velocity graph corresponding to piston movement. This graph is generally only valid for motion in one direction in a given region. In other directions or regions, the value of for instance V_piston may be different.

If the valve controller operates the valve unit to apply pressurized gas to one of the inlets and to connect the other one to the exhaust, the piston will first not react for a given delay time t_dei- This time is related to the inevitable delay times that are associated with pressure release and response times of mechanical components for instance due to friction and other influences. After this time delay, the piston will start to accelerate towards velocity V_piston during an acceleration time t_acc. It is assumed that this velocity ramp-up is linear, which allows for easy calculation of the increase in (virtual) position during acceleration. Subsequently, the piston will move at the steady-state velocity V_piston during a time t_iin. Shutting off the corresponding valves to force a stop of piston 2 will not immediately result in a standstill of piston 2. Instead, piston 2 will continue to move during a time t_dec in which it will decelerate. It is assumed that this velocity ramp-down is also linear, again allowing easy computation of the (virtual) position increase during this period. There even may be additional delay involved with shutting down the release and other influences such as friction. If needed, these effects can be accounted for by a separate calibration parameter. This parameter could be included in the calibration similar to the other parameters.

The valve controller can only control the duration of the period in which the valve unit is in the extension or retraction connection state. This period corresponds to the time t_dei+t_acc+ti_iri in figure 3. In case, t_dei, t_acc, and t_dec cannot be neglected compared to t_iin, they must be present in the calibration. Given these parameters, the position of piston 2 in figure 2 can be traced without exactly knowing or measuring the position itself.

Figure 4 shows a flow-chart corresponding to a movement from a start position to a target position of the piston. The process starts with step Sl with the piston being at a known position, for instance fully retracted. In this case, the virtual position corresponds to the true position and can be denoted as V_pos=start. Then, the valve unit receives an instruction to move from this position to an intermediate position. This position does generally not correspond to any position detectors which may be present. Therefore, there are no means by which it can be verified whether the piston has actually reached that position. Instead, a calibration is used which transforms the control signal to the valve unit, e.g. the activation time, into a displacement of the piston. The activation time is the time the valves are operated to be in the extension (or retraction) connection state as a result of which the piston will move. Using the calibration, the effect of putting the valve unit in the extension connection state can be predicted. The calibration describes the motion characteristics of the piston during piston displacement. A typical example of motion characteristics was shown in figure 3.

Referring back to figure 3, given a known start position Pl and target position P2, the system must compute the activation time needed to achieve this motion. For that, it calculates the spatial difference between the two positions in step S2. Then it uses this difference and the calibration Cl to calculate the activation time in step S3. For the motion characteristics in figure 3, it is assumed that t_dei, t_acc, tdec and V_piston are known. Using these data, the required activation time t_act can be computed. Subsequently in step S4, the valve unit is operated for the duration of the activation time t_act ending with the piston coming to a standstill in step S5.

The virtual position can be updated using a counter or timer value, see figure 5. Such a counter or timer is started in step S6 after issuing the relevant instruction to the valve unit in S4. However, the updating process is not started until a period of t_dei has passed, see step S7. This corresponds to the time that the piston does not move. Then, the virtual position V_pos,_s9 is updated during the period that the piston is accelerating. It should be noted that V_pos,_s9 corresponds to the virtual position during the execution of step S9. Given a predefined velocity profile, see for instance figure 3, the traversed distance can be calculated which can be used to update the virtual position, see S8. This update is continued until a time of t_acc+t_dei has passed, see step S9. After this time, the piston is assumed to move at a constant velocity V_piston- Again, the virtual position can be updated using the elapsed time, see step SlO, until the activation time t_act has passed, see step SIl. This time corresponds to tdei+tacc+tim. Then, the valve unit is switched to the lock connection state. However, the piston will first decelerate for a period of t_dec- Again, assuming a predefined velocity profile, the virtual position can be updated, see S12. After the delay time has passed, the virtual updating process will stop, see steps sl3 and S14.

It should be noted that further delay times can be introduced for instance corresponding to the time it takes for the system to go from the extension connection state to the locking connection state. Any parameter may be introduced as long as this parameter is in the calibration.

Instead of updating during the displacement of the piston, the virtual position can also be updated only once, for instance after the piston has come to a standstill. This reduces the complexity of the algorithms somewhat, but information regarding the temporal position of the piston in between the target and start positions is lost.

Normally, the motion characteristics of the piston are not symmetric. For instance, the effective frictional force encountered by the piston is different for the extending and retracting strokes. Furthermore, this force may be different depending on the area or region in the stroke range the piston is located in. Reasons for asymmetry are a changing load of the piston depending on the position of the piston, different effective surfaces of the piston possible combined with different pressures, different speed regulator settings, different friction in attached mechanism depending on position mechanism also in relation with load, different friction in cylinder, difference in wear out of cylinder which influence friction and leakage, difference in friction of cylinder seals depending of position and direction, difference in distance of regions length, difference in sensor activation depending of direction of movement, etc. To account for these asymmetries it is advantageous to use a bi-directional calibration. Bi-directional implicates that the relevant calibration parameters depend on the actual movement of the piston.

Furthermore, it is convenient to use different bi- directional calibrations for different regions. A possible division of the total stroke range of the cylinder is depicted in figure 6, which illustrates a pneumatic cylinder 1 in which the total stroke range is divided in two regions. To get feedback information from the physical (true) position of piston 2, a piston is used that includes a magnetic member 24 as well as a reed switch 25. Reed switch 25 changes its state when the magnetic member 24 passes one of its flanks 26, 26' . In figure 6 the total stroke range 27 is divided in an inner region 28 and an outer region 29. The boundary between these regions is defined by flank 26' of reed switch 25. Consequently, inner region 28 runs from one edge 30 of stroke range 27 to flank position 26', whereas outer region 29 runs from flank position 26' towards the other edge 31 of stroke range 27.

Inner region 28 and outer region 29 are each characterized by a different bi-directional calculation. This would for instance result in four different values for t_dei, two for each region corresponding to the extending and retracting stroke. To update the virtual position, as well as to calculate the needed activation time, the different calibrations must be used. For instance if the start position is in the inner region and the target region is in the outer region, one needs the calibration parameters corresponding to an extending stroke in the first region to compute the activation time needed to move from the start position to the boundary between the regions, and the calibration parameters corresponding to an extending stroke in the second region to compute the activation time needed to move from the boundary position to the target position. It should be noted that in this case, i.e. a single stroke, one does not need to take into account the deceleration delay for the first region.

It may be possible that during the movement from start to target position, the piston crosses a position that is detectable by a position detector such as the reed switch in figure 6. In that case, it is convenient to set the virtual position equal to that known position at the time the piston is actually at that position. It may further be advantageous to re-calculate the activation time that is still required to move to the target position. Because no continuous feedback is available, it is advantageous to express the various positions relative to the total stroke range or region length. For instance, in figure 6, the fully retracted position is denoted as "0", whereas the fully extended position is denoted as "S", "S" being a dimensionless number. In effect, the positions are divided by the total stroke range. In this way, the system does not need to know the exact value of the total stroke range or the exact value of the target position, only relative positions are required, e.g. 50 % of the total stroke range. Similarly, distances could be expressed with reference to the length of a region, 50 % of region.

Up till now, the calibration parameters have been assumed known. For now, it is assumed that the calibration parameters are determined using dedicated calibration routines. An example of such a routine is given in figures 7A-7C. In figure 7A, the piston is instructed to move from a known position in the inner region to a target position in the outer region. The time needed to cross the detection position is recorded. This time corresponds to tdei+tacc+tim. In figure 7B, the same target position is set but then assuming a sequence of strokes. In both cases, it is verified that the piston actually crosses the boundary between the different regions, and the corresponding time is measured. Figure 7C also depicts a sequence of strokes but then using a different activation time t_act per stroke. In all three cases, it is verified that piston crosses the position detector while moving at the steady-state velocity. This can be achieved by making sure that this position is crossed just before expiry of the relevant activation time. In figure 7A, the time measured to move from the start position to the position detected by the position detector in a single stroke is given by:

1

where ti_in_i corresponds to the time of constant speed for the 5 single stroke motion. From this, the distance can be calculated using:

EQ 2: V_plston/2 X t_; (t me a s 1 -tdel-t :) X V_p p_liston

10 which distance should correspond to the distance between the start position and the relevant position corresponding to the position detector. When the position is detected, the valve unit is switched to the lock connection state immediately. The piston will move a bit further due to the

15 deceleration delay, t_dec-

In figure 7B, the same distance is covered using multiple strokes with measured time t_meas_₂, whereas figure 7C depicts the situation with measured time t_meas_3 • These times correspond with the corresponding activation times and are

20 therefore known.

For figure 7B the traversed distance can be calculated using :

EQ 3: V_plston/2 X 2 X t_aCc + V_plston/2 X 1 X tdec + (t_meas_2-

Z D TL del TL ace / X ^piston ^"*^" \ T-meas_2 ' ^"^ ctel tacc/ X ^piston

For figure 7C it can be found using:

EQ 4: V_piston/2 x 3 x t_aCc + V_pistOn/2 x 2 x t_dec + 2 x

•-* U ( ^meas_3 TL ctel ^"^acc/ X ^piston ^"*^" \ ^meas_3 ' ^"^ del TLacc/ X ^piston

Similar equations can be found for an even higher number of steps. Equating the equations above allows for the elimination of V_piston- Accordingly, the remaining parameters can be determined. However, t_dei could also be determined by using an extrapolation technique in which the activation time per stroke is gradually decreased up to the moment no visible displacement of the piston is noticeable. It could also be determined by first moving the piston just beyond a position corresponding to a position detector and to then direct the piston backwards. In this case, the activation time is increased until the piston reaches the aforementioned position. Similarly, t_dec could also be determined by moving the piston from one of the extremes of the stroke range to a position associated with the position detector. The piston should first be activated during an activation time which is sufficient for the piston to reach the position detector at nominal velocity. Then the activation time should be reduced. By plotting the measured time required for the movement versus the activation time, tdec can be extracted. In such a plot, three distinct regions can be distinguished. In a first region, the activation time is high enough for the piston to reach the desired position with nominal velocity. In a second region, the piston will reach the desired position after the system has been deactivated, i.e. during tdec. In a third region, the piston will not reach the desired position because it has come to a standstill before the desired position.

Up till now, Vp_lston has been assumed to be known. This parameter can however be determined using a measurement between two known positions, for instance between an extreme of the stroke range and a position corresponding to a position detector. The time required for moving the piston from the extreme of the stroke range to the position equals tdei+tacc+tiin. This value can be measured by timing the required activation time. Because t_dei and t_acc are known, t_iin can be calculated. From this value and using equation 2, Vp_lston can be computed. The absolute value would then be equal to L/ (ti_in+0.5t_aCc) whereas the relative value would for instance be 1/ (ti_in+0.5t_aCc) • In the latter case, distances are computed with respect to region length L.

Using the approach described above, the relevant calibration parameters for one direction in one region can be determined. The calibrations for other regions and or directions may be calculated in a similar way. For some applications, the accuracy and or precise knowledge of the virtual position is less important. In these cases, it might not be required to know the various parameters, e.g. t_acc, t_dei, etc. Instead, a single correction factor t_x can be used. This factor describes the difference between covering a given distance in a single stroke and in multiple strokes.

The distance L covered in a single stroke can be written as :

EQ 5 : L = t_meas_i x V_pistOn

wherein V_piston represents a fictitious average speed the piston would have if it would move only during t_meas_i • It should be noted that this speed incorporates the influences of the calibration parameters described so far, e.g. t_dei, tacc, tdec- Furthermore, L typically corresponds to the length of the region.

The same distance covered in n strokes can be expressed as :

EQ 6 : L = n x (t_meas_n-t_x) x V_pistOn wherein t_meas_n represents the corresponding activation time, and wherein it is assumed that the piston moves at a similar speed. Equation 6 indicates that due to the fact that multiple strokes are used, the effective activation time is decreased by t_x.

Similarly to figures 7A-7C, the value of t_x can be determined as a function of the number of steps needed to traverse a given distance. By combining equations 5 and 6, a correlation can be found between t_meas_n and t_x. If the piston should move between two relative positions, e.g. 10% and 20% of a given region length, the relative distance to be traversed can be calculated, e.g. 20%-10%=10%, which corresponds to n=10. Using this value, a value for t_x can be found. If needed, the correlation found using equations 5 and 6 can be interpolated or extrapolated.

The above mentioned relative speed can be used to determine the required activation time:

EQ 7: tact = t_x + (0.2 x L - 0.1 x L) / (L/t_meas_i)

wherein L can be eliminated. During the motion of the piston starting from V_pOs, start, the virtual position can be updated for instance using:

EQ 8 : V_{po s} = V_{po s} , _start + ( t ime-t_x ) x ( L /t_meas_i )

wherein the position is updated after a time t_x has passed. If relative positions are used, e.g. V_pos/L, length L can be eliminated and therefore need not be known. It should further be noted this method can be applied for each region and direction individually.

During the movements of the piston, it may happen that one of the displacements corresponds to a calibration displacement as for instance presented in figures 7A-7C. In that case, the calibration can be re-calculated based on the newly measured time. For instance, if a piston is to traverse a region in a single stroke, at least one of a retracting stroke and an extending stroke is performed, which corresponds to the displacement in figure 7A. The measured time can be used together with previously determined times, e.g. t_meas_2 and t_meas_₃, to recalculate the calibration parameters. However, the calibration displacements may also be part of the motion sequence. In that case, the calibration is performed on a regular basis.

It may also prove necessary to perform the calibrations in a particular sequence. For instance, in order to calibrate the system for retracting strokes in the inner region, the piston needs to move between two known positions. Having first done the extending stroke calibration in the inner region, the piston can be moved accurately to the detector position, which is a first known position. The other position, for instance, a position in between an end of the stroke range and the detector position is generally not known, i.e. it cannot be determined at which time the piston reaches this position. However, this position or time can be determined by using an extending stroke starting from this position to the detector position. Above, the invention has been described by reference to a pneumatic system using a pneumatic cylinder. However, as already stated, the disclosed principles could equally be applied to hydraulic systems.

Furthermore, although the invention has been described using various embodiments thereof, it should be obvious to the skilled person in the art that various modifications are possible without exceeding the scope of the present invention that is defined by the appending claims. For instance, the calibration of one cylinder could depend on the position of other cylinders. This is particularly important in case the cylinders are mechanically coupled, such as cylinders corresponding to multiple joints in an arm of a mannequin robot. A relationship between the position of one cylinder and the calibration of another could be cast into mathematical formulae which comprise one or more constants. These constants can then be determined during the calibration and or during regular motion of the cylinders.

Furthermore, the system could be extended with controlling the pressure available at the various stages of the systems, such as the inlets or valves. This provides the possibility to accelerate and decelerate smoothly. It also gives the possibility to change the pressure smoothly. This can be necessary if the pressure changes due to external loads. With pressure control, also speed regulation is made possible. The speed of the piston could for instance be controlled by proportional speed regulators and or valves. The system can be further improved by adding more sensors and by using the information from these sensors during the calibration and or during the regular motion of the cylinder, for instance for equating the virtual position to a measured position. An example of such a detector could be another end of a reed switch. This applies if the corresponding region of the cylinder is longer than the detection region of the reed switch.

Additionally, the system may keep track of a history of calibration data. Should there be large deviations between newly determined or measured values and corresponding stored values, the system may decide to redo part of the calibration or to ignore the newly determined parameter completely . Furthermore, the present invention has been disclosed under reference to a linear cylinder. However, the principles outlined in this description could equally be applied to a rotary cylinder. Within the context of the present invention, the extending and retracting strokes should be interpreted as relating to rotations in opposite directions .

Claims

1. A method for controlling piston displacement of a cylinder, said cylinder having a first and second inlet corresponding to a retracting and extending stroke of the piston of the cylinder, respectively, said method comprising the steps of: providing a source of pressurized medium;

- providing an exhaust; - providing a valve unit for connecting the source of pressurized medium, the exhaust and the first and second inlet;

- controlling the valve unit to switch to a retraction connection state for achieving said retracting stroke, in which the first inlet is connected to the source of pressurized medium and the second inlet is connected to the exhaust, or to an extension connection state for achieving said extending stroke, in which the second inlet is connected to the source of pressurized medium and the first inlet is connected to the exhaust; characterized by locking said piston by switching the valve unit to a lock connection state in which both the first and second inlet are connected to the source of pressurized medium.

2. The method according to claim 1, wherein the cylinder comprises a pneumatic cylinder, and wherein the source of pressurized medium comprises a source of pressurized gas.

3. The method according to claim 1, wherein the cylinder comprises a hydraulic cylinder, and wherein the source of pressurized medium comprises a source of pressurized liquid.

4. The method according to any of the claims 1-3, wherein during the locking of said piston, medium is prevented from exiting the first and second inlet by using at least one non-return valve.

5. The method according to any of the claims 1-4, further comprising regulating the speed at which the medium exits the first or second inlet for controlling the speed of said piston during piston displacement.

6. The method according to any of the preceding claims, further comprising regulating the pressure of the pressurized medium that is supplied to the first and or second inlet.

7. The method according to any of the preceding claims, further comprising determining whether the piston is in a predefined region in a stroke range of said cylinder.

8. The method according to any of the preceding claims, further comprising determining whether the piston is at a predefined position in a stroke range of said cylinder.

9. The method according to any of the preceding claims, wherein said controlling of piston displacement comprises applying the exhaust to the first or second inlet for a predetermined amount of activation time to displace the piston from a start position to a target position, said method further comprising the steps of: providing a calibration with calibration parameters of the cylinder, said calibration describing the motion characteristics of said piston during piston displacement; computing a virtual position corresponding to the start position of the piston; calculating a difference between the virtual position and the target position; - determining an activation time for the cylinder based on the calculated difference and the calibration;

- controlling the cylinder using the determined activation time;

- updating the virtual position using the calibration and the amount of elapsed activation time during said piston displacement .

10. The method according to claim 9, wherein the start and target positions are expressed relative to one of the stroke range and the region length, and wherein the calibration parameters comprise the amount of activation time needed for piston displacement between two known and or detectable positions .

11. The method according to claim 10, wherein said calibration further comprises at least one or a combination of the following calibration parameters: a start delay time corresponding to a delay in time between controlling the valve unit and resulting motion of the piston; an acceleration time corresponding to the amount of time needed for acceleration from a standstill of said piston to a corresponding steady-state velocity; and

- a deceleration time corresponding to the amount of time needed for deceleration from a corresponding steady-state velocity of said piston to a standstill.

12. The method according to claim 11, wherein providing the calibration comprises: measuring the time needed to complete an extending and or a retracting stroke between two known or detectable positions in a single stroke; measuring the time needed to complete an extending and or a retracting stroke between the same two known or detectable positions in a sequence of strokes;

- comparing the single and sequence of strokes measurement for extraction of at least one calibration parameter or a combination thereof as defined in claim 11; and storing the extraction result in the calibration.

13. The method according to any of the claims 10-12, wherein providing the calibration comprises:

- measuring the time needed to complete an extending and or a retracting stroke between two known or detectable positions in a single stroke; and - storing at least one of said measured time, a speed of the piston determined using this measured time, and a speed of the piston determined using this measured time relative to the total stroke range or a length of a region, as a calibration parameter in the calibration.

14. The method according to claim 12 or 13, comprising determining a relevant calibration parameter and storing this parameter if the displacement of said piston between the start and target position comprises the single stroke as defined in claim 12 or 13 or the sequence of strokes as defined in claim 12.

15. The method according to any of the claims 9-14, in so far as depending on claim 8, further comprising setting the virtual position equal to the predefined position when the piston is at said predefined position.

16. The method according to any of the claims 9-15, wherein the calibration comprises separate calibration parameters for the retracting and extending strokes thereby forming a bi-directional calibration, wherein said updating of the virtual position and said determining of the activation time is done in dependence of the motion direction of the piston using the corresponding bi-directional calibration.

17. The method according to any of the claims 9-16, wherein the stroke range of the cylinder is divided in at least two regions, each region being characterized by a corresponding bi-directional calibration, wherein said updating of the virtual position and said determining of the activation time is done in dependence of the motion direction of the piston and the region in which the piston is positioned using a corresponding bi-directional calibration.

18. The method according to claim 17, wherein said updating of the virtual position and said determining of the activation time is further done in dependence of any region the piston may traverse during said piston displacement.

19. The method according to claim 17 or 18, in so far as depending on claim 7 and or 8, wherein a position and or region detector is placed at the boundary between adjacent regions .

20. System comprising:

- a cylinder having a first inlet and a second inlet corresponding to an retracting and extending stroke of a piston of the cylinder, respectively; - a source of pressurized medium; an exhaust;

- a valve unit for connecting the source of pressurized medium, the exhaust and the first and second inlet; - a valve controller for controlling the valve unit to switch between connection states; wherein the system can be operable in: an retraction connection state for achieving said retracting stroke in which the first inlet is connected to the source of pressurized medium and the second inlet is connected to the exhaust, and

- an extension connection state for achieving said extending stroke in which the second inlet is connected to the source of pressurized medium and the first inlet is connected to the exhaust; characterized in that the system is further operable in a lock connection state for locking the position of said piston in which both the first and second inlet are connected to the source of pressurized medium.

21. The system according to claim 20, wherein the cylinder comprises a pneumatic cylinder, and wherein the source of pressurized medium comprises a source of pressurized gas.

22. The system according to claim 20, wherein the cylinder comprises a hydraulic cylinder, and wherein the source of pressurized medium comprises a source of pressurized liquid.

23. The system according to any of the claims 20-22, further comprising at least one non-return valve in between the source of pressurized medium and the valve unit.

24. The system according to any of the claims 20-23, further comprising a speed regulator in between the first and or second inlet and the valve unit .

25. The system according to any of the claims 20-24, further comprising a controllable pressure regulator to regulate the pressure of the pressurized medium which is applied to the first and or second inlet.

26. The system according to any of the claims 20-25, wherein the valve unit comprises: a first valve arranged to connect the first inlet to the source of pressurized medium in the retraction and lock connection state and to connect the first inlet to the exhaust in the extension connection state; a second valve arranged to connect the second inlet to the source of pressurized medium in the extension and lock connection state and to connect the second inlet to the exhaust in the retraction connection state.

27. The system according to any of the claims 20-26, further comprising a region detector to detect whether the piston is within a predefined region in a stroke range of said cylinder.

28. The system according to any of the claims 20-27, further comprising a position detector to detect whether the piston is at a predefined position in a stroke range of said cylinder .

29. The system according to claim 28, wherein the position detector comprises a medium sensor connected to the first and or second inlet, which is arranged to measure the corresponding medium volume output.

30. The system according to any of the claims 27-29, wherein the piston is provided with a magnetic member and wherein the region and or position detector comprises a reed switch coupled to the cylinder to detect said magnetic member .

31. The system according to any of the claims 27-30, wherein the valve controller comprises:

- an input unit for receiving instructions to displace the piston from a start position to a target position by applying the exhaust to the first or second inlet for a predetermined amount of activation time;

- a calibration memory for storing calibration parameters of a calibration, said calibration describing the motion characteristics of said piston during piston displacement; a virtual position unit arranged for computing a virtual position corresponding to the start position of the piston;

- a virtual position memory for storing the virtual position; an activation time calculate unit for calculating the needed activation time based on a difference between the virtual position and the target position and the stored calibration; wherein the virtual position unit is further arranged to update the stored virtual position using the stored calibration and the amount of elapsed activation time during said piston displacement.

32. The system according to claim 31, further comprising a calibration unit to determine at least one of the calibration parameters defined in claim 10 and or 11, wherein the calibration unit is arranged to determine said at least one calibration parameter using the method as described in claim 12 and or 13, respectively.

33. The system according to claim 31 or 32, wherein the calibration unit is arranged to update the stored calibration if the displacement of said piston between the start and target position comprises the single stroke as defined in claim 12 or 13 or the sequence of strokes as defined in claim 12.

34. Mannequin robot comprising a system as defined in any of the claims 20-33.