Integration of artificial intelligence in robotic vehicles: A bibliometric analysis

1 Introduction

It was once a myth that robots would coexist with humans, but now it is a fact because of growing technologies. The word robot is taken from the Czech language, meaning worker. Robots are invented to support humans by sharing their efforts [1]. George Devol introduced the first modern robot with patent rights in 1954, but it was a commercial failure. To address these issues and introduce the robot for die casting in 1961, Joseph Engelberger and George Devol collaborated. Robots have begun to acquire popularity and spread across many industries, and with the aid of enabling technologies such as sensors, robots are expanding into new sectors [2].

Robots are divided into six groups based on their functions, including automated guided vehicles (AGVs), autonomous mobile robots (AMRs), humanoids, articulated robots, cobots, and hybrids.

AGV robots are mainly designed for transportation in the factory to transport materials and goods. These robots are capable of working in any situation and require a path to follow [3]. Karl is one of the AGVs for carrying materials with its arm. It follows the path and identifies the objects using radio frequency identification [4]. AMRs are designed to function in industries including manufacturing, agriculture, and medicine. There is no need for a predefined course because they are capable of making their own decisions and assessing their environment with the aid of cameras and sensors. It can detect the human head, adjust its course, and select the precise package [5]. The best example of this type of device using sensors, such as ultrasonic sensors, is a robot vacuum cleaner. It can more accurately clean the floor and can foresee barriers such as rugs and sandals so that it can adjust its course. It cleans more thoroughly than a regular vacuum cleaner [6]. According to the specifications, robot cars fall within the AMR and AGV. Humanoid robots are designed to make interaction with humans and share work with humans. These come with an advanced version of AMR that can sense surroundings and take decisions and looks like and walks like a human. The main purpose is to reduce the risk of humans in the hazard situation; instead of humans these robots can work in such situations, as caretakers for the disabled and aged people and share household work as a supporter [7]. Sophia is the first humanoid robot that can walk and talk like a human, sense its environment, and respond appropriately. The creators of Sophia garnered a lot of praise, and after appearing in numerous worldwide interviews, Sophia was granted citizenship – the first non-human to do so [8]. To assist people, articulated robots are built with rotatory joints that have a minimum of two and a maximum of more than ten. These robots, which are also known as human arms, add arms to people to share work and shorten task times. These are made with the use of electric motors [9]. Examples of this kind include robot arms, which are utilized in the automobile sector for a variety of tasks while the vehicle is being built. Robotic arms that can weld with both large and low payloads are available for the car body [10]. Cobots are a kind of robot family that is designed to help them in working for collaborative work with humans. These robots work beside humans [11]. With the change in the trend of technologies such as the internet of things and artificial intelligence (AI), robots started integrating them.

Integration of AI in robots gives brains to the robots that can observe an object and take a decision without human involvement. AI makes a robot autonomous with the help of machine learning [12]. One example of AI integration in robotics is Alexa: a voice assistant that helps individuals [13]. The challenges of integrating AI into robots are data security [14], algorithm selection [15], and so on. Section 2 provides preliminary data analysis, which includes preliminary data analysis of the initial search results, and Section 3 describes the bibliometric analysis, which includes geographic region analysis, network analysis, and statistical analysis of publication citation.

2 Preliminary data

Bibliometric analysis, which examines journal articles, conference papers, book chapters, books, and brief surveys, aids academics in identifying research shortages in particular fields [16]. This research article is a bibliometric analysis of the integration of AI in robotic vehicles. This analysis extracts data and clears all queries from the database of Scopus. The database of citations is with a large variety of field domains such as agriculture, medicine, biological sciences, engineering, computer science, social science, and science. The data are yielded from the database of Scopus on 18 August 2022. The search option for the keyword combinations is “Artificial intelligence” AND “Robotic vehicles” for the duration from 2015 to 2022. The top ten keywords in this analysis are itemized in Table 1 with the number of publications.

2.2 Preliminary data analysis

Figure 1 illustrates the number of publications per year. The most publications are made in the year 2021 with 252, and as publications are growing each year, this is an excellent field for further research. Figure 2 describes the top ten authors and their publications. Cowlagi R. V. has more publications of 17, and the remaining 9 authors have 6 publications.

Figure 1

Documents published per year.

Figure 2

Top ten authors with publications.

Figure 3 describes the top ten countries with the number of documents. The United States has the most number of publications with 290 and China and India are in the second and third place with 247 and 122 publications. Figure 4 describes the type of document with the number of publications. Articles are the document type with a large number of publications, around 593 with a percentage of 49.6%.

Figure 3

Top ten countries with publications.

Figure 4

Type of document with the number of publications.

Figure 5 shows the document per year by source. IEEE Access has publications more than 15. Figure 6 shows documents by subject area. Engineering subject domain has most of the publications with a percentage of 31.4% and Computer Science has the second most publications with a percentage of 29.7%.

Figure 5

Document per year by source.

Figure 6

Document with subject area.

Figure 7 describes funding agencies with the number of publications. National Natural Science Foundation of China is the funding sponsor for most of the publications with 114.

Figure 7

Funding agencies with a number of publications.

3 Bibliometric analysis

Bibliometric analyses of the AI integration in robotic vehicles are listed subsequently.

3.1 Territorial analysis

The top three countries in the publication are the United States, China, and India. More publications are made by the United States with 290 publications. Figure 8 describes the locations of the top ten countries where the research on AI integration in robotic vehicles is carried out. Figure 8 was obtained with the help of the open-source platform gpsvisualizer.com with the GPS visualizer tools.

Figure 8

Top ten geographical locations worldwide.

3.2 Visualization with network

This network visualization is made with the help of open-source tools such as Gephi, VOSviewer, ScienceScape, and Wordcloud. Figure 9 describes the congregation of year and publication title of AI integration into robotic vehicles from the database of Scopus. More number of publications are published in the year 2021 with 252 publications. Figure 10 describes the congregation of author and author keywords together appearing on the paper.

Figure 9

Congregation of year and publication title from the database of Scopus.

Figure 10

Congregation of authors and author keywords on the same paper.

Figure 11 shows the congregation of keywords that are occurring together in a paper. The interrelations between keywords that are coappearing in the research are linked together. Autonomous vehicles, robotics, vehicles, agricultural robots, and navigation are the most coappearing keywords.

Figure 11

Congregation of co-appearing keywords from the database of Scopus.

Figure 12 shows the diagram of the top authors, keywords, and journals and the interconnections between them like a Sankey diagram. The first column describes the top author’s, the second column describes their top keywords, and the third column describes their journals. Figure 13 is a table that describes Figure 12. The first column is with the main authors, the second column is with the main keywords, and the third column is with the main journals. The machine learning keyword is used in 48 papers.

Figure 12

Sankey diagram of top authors, keywords, and journals.

Figure 13

The table representing Figure 12.

Figure 14 shows that the words used in the publications are visualized in the word cloud. The keywords which are frequently used are shown in large font sizes. The words “control,” “surface,” “systems,” “usvs,” “vehicles,” and “usv” are the most appealing words in the highest cited paper.

Figure 14

Word cloud of the highest cited paper.

4 Recommendations and scope

Robotic vehicle AI integration is a rapidly expanding field of study that is appealing to both novice researchers and seasoned researchers. Robots have a lot of potential in the future because humans cannot work in hazardous situations, and in such life-risky situations robots help us to do that work. The fact that there are more publications in this field every year indicates that there is a lot of research gap to do analysis. Robots are increasingly incorporating new technologies to make them more precise and appropriate, but as technology advances, it may also be possible to identify both the benefits and drawbacks of a certain technology. Then, we can work on the merits to put them into practice to discover new technology, as well as on the demerits to find a remedy. This cycle keeps going, allowing us to work in this direction.

AI makes a robot take its own decisions and work accordingly but there is a chance of malfunctioning. The future direction of AI integration in robotic vehicles is to work on security to avoid vulnerabilities and malfunction.

5 Conclusion

The bibliometric analysis of the integration of AI in robotic vehicles is based on the data taken from the Scopus database; robotics is the most used keyword, and English is the language used in more publications. Articles are the type of documents that have more publications. The United States is the country in which this domain research is mainly going on. The visualization analysis is done with free source tools such as Gephi, VOS viewer, GPS visualizer, ScienceScape, and word cloud. AI plays a crucial role in robots nowadays to reduce the involvement of humans to control them instead the robots can take their own decisions with the help of the sensors by sensing the surrounding environment. This bibliometric analysis helps researchers to find research gaps in the integration of AI in robotic vehicles.