Tesla app size: 760 MB, what's inside?

#C++ Diary 📒 #Inheritance in C++ is a mechanism in Object-Oriented Programming (OOP) where a class (derived class) can acquire the properties and behaviors of another class (base class). This helps us achieve: • Code reusability → Common functionality written once can be reused across multiple modules. • Hierarchy representation → Models real-world “is-a” relationships (e.g., Car is a Vehicle). • Extensibility → We can add new specialized behaviors without rewriting everything. ⸻ 🔹 Types of Inheritance in C++ 1. Single inheritance → One base, one derived. 2. Multiple inheritance → One class inherits from multiple base classes. 3. Multilevel inheritance → Chain of inheritance (e.g., Vehicle → Car → ElectricCar). 4. Hierarchical inheritance → Multiple derived classes from one base class. 5. Hybrid inheritance → Combination of the above. ⸻ 🔹 Automotive Real-Time Example Imagine we are working in automotive ECU (Electronic Control Unit) software. We need to design a test framework for different types of vehicles (ICE cars, Hybrid, EVs). We can model this hierarchy using inheritance: —-$$$Program$$$—- #include <iostream> using namespace std; // Base Class class Vehicle { public: void start() { cout << "Vehicle is starting...\n"; } void stop() { cout << "Vehicle is stopping...\n"; } }; // Derived Class - ICE Car class ICECar : public Vehicle { public: void refuel() { cout << "Refueling petrol/diesel...\n"; } }; // Derived Class - Electric Car class ElectricCar : public Vehicle { public: void recharge() { cout << "Recharging battery...\n"; } }; // Derived Class - Hybrid Car class HybridCar : public Vehicle { public: void switchMode(string mode) { cout << "Switching to " << mode << " mode...\n"; } }; int main() { ICECar benz; benz.start(); // Inherited behavior benz.refuel(); // ICE-specific behavior benz.stop(); ElectricCar tesla; tesla.start(); tesla.recharge(); // EV-specific behavior tesla.stop(); HybridCar toyota; toyota.start(); toyota.switchMode("EV"); // Hybrid-specific behavior toyota.stop(); } In real automotive projects (like at Bosch, Daimler, or Mobileye), inheritance is heavily used to model ECU diagnostics, vehicle components, and communication stacks. Example use case: • Base class → DiagnosticService (common UDS diagnostic request handling). • Derived classes → ReadDTC, ClearDTC, ReadDataByIdentifier, etc. • Each derived class inherits the standard behavior (session handling, security unlocking) from DiagnosticService but implements service-specific logic. Another example: Network ECUs • Base class: ECU (common CAN/Ethernet communication methods). • Derived class: PowertrainECU, ADASCameraECU, InverterECU. • These inherit base communication but extend for their specialized needs (e.g., torque control for Inverter ECU, lane detection for Camera ECU).

🔄 AUTOSAR CAN State Machine: The Invisible Guardian of Your Car's Communication Think CAN drivers are simple? Think again. Behind every message your ECU sends lies a sophisticated state machine that ensures reliability, safety, and determinism. Here's why it matters: 🏗️ The Two-Layer Architecture Module Level (Simple): CAN_UNINIT → Power-up default CAN_READY → After Can_Init() Called only once by ECU State Manager Controller Level (Complex): UNINIT → Hardware in reset STOPPED → Initialized but silent STARTED → Fully operational SLEEP → Low-power with wake-up 🎯 Key AUTOSAR Requirements Safety First Design: ❌ Invalid transitions? → CAN_E_TRANSITION error ❌ Double initialization? → Development error ✅ Timeout monitoring → Uses GetCounterValue() ✅ Non-blocking → Can_MainFunction_Mode polls status Bus-Off Handling: Automatic STARTED → STOPPED transition Cancel pending messages immediately No auto-recovery - Upper layers control restart Notify CanIf via CanIf_ControllerBusOff 🚗 Real-World Impact Brake-by-Wire Example: UNINIT → Can_Init() → STOPPED System ready → CAN_CS_STARTED → Operational Bus-off detected → Auto STOPPED + Diagnostics Safe recovery initiated by application Energy Saving: Running: STARTED → SLEEP (hardware or emulated) Wake-up: CAN message → STOPPED → STARTED Critical for electric vehicle battery life 💡 Why This Architecture Matters Deterministic Behavior: No race conditions in state transitions Clear timeout handling prevents system locks Predictable error recovery paths ISO 26262 Compliance: Required for safety-critical systems Every transition is verifiable and testable Error handling built into state machine 🔧 Developer Best Practices c // Non-blocking state transition with timeout Can_SetControllerMode(CAN_CS_STARTED, timeout_ms); // Bus-off recovery handled by upper layers void CanIf_ControllerBusOff(uint8 ControllerId) { Dem_ReportError(CAN_BUS_OFF_EVENT); ScheduleRecovery(ControllerId); // Application decides when to restart } 📊 Common Pitfalls to Avoid ❌ Blocking waits for hardware state changes ❌ Assuming STARTED state without verification ❌ Missing bus-off handling - system won't recover ❌ Ignoring wake-up events - power management fails 🎯 Key Takeaways Separation of concerns: Module vs controller states Hardware events trigger automatic transitions Timeout monitoring is mandatory, not optional Upper layers own recovery - driver only reports events Sleep mode support varies - emulation may be required This state machine is why your car's CAN communication remains reliable even under fault conditions. The strict transition rules prevent unpredictable behavior that could lead to safety-critical failures. 🤔 Question: What's the most challenging state management issue you've faced in automotive systems? #AUTOSAR #CANBus #EmbeddedSystems #Automotive #StateMachines #ECU #SafetyCritical #ISO26262 #EmbeddedSoftware #CarTech

Ferrari just deleted 50% of their apps. Their fans are celebrating. Microsoft added AI to everything. Their users are... confused. One company cut ruthlessly. One company added endlessly. Guess which product experience actually improved? When Ferrari team principal Frédéric Vasseur greenlit their new AI-powered fan experience with IBM, he knew it meant killing half their legacy stack. But he understood something most product teams miss: AI isn't a feature you bolt on. It's a foundation you build on. The result? Real F1 fans said the new app made them feel closer to the team, smarter about the race, and part of the pit wall. Meanwhile, Microsoft's "VibeWorking" is like putting a Tesla battery into a diesel tractor. I broke down the full story—and the 5-step framework to kill your own product bloat—in my latest newsletter: https://lnkd.in/evnrimbW The best products aren't built by adding more. They're built by cutting everything that slows you down. What would YOU cut from your product if you could start over? 👇

The energy and conversations at #SFTechWeek have solidified a powerful question that hangs over the entire industry: The PC had Windows. The smartphone has Android. What is the standard operating system for the coming robotic age? The answer is: it doesn’t exist yet. ⚡️ That is the multi-trillion dollar void Hillium was founded to fill. We are building HilliumOS: the first open platform for Kinetic AI. Our mission is to do for robotics what Android did for mobile—create a universal standard—but built from the ground up with a core of immutable safety ecosystem. The platform for the next great technological revolution isn't coming. It's being built. This past week in SF was a powerful confirmation. The future is closer than we think. #KineticAI #HilliumOS #Robotics #VentureCapital #DeepTech #Innovation #Platform

Robotics (a bit of a geek post). Is Apple about to lose its advantage in robotics? I've loved the unified memory in Apple, as it was doing away with the separation between RAM and VRAM that's present in the Windows/Linux ecosystems, for far less USD, so it has been a great value from an AI perspective. Apple though is becoming of very limited use in robotics, when it comes to training (still great for inference). None of the Issac/Omniverse like tools => that's only Nvidia. BUT, the recently announced DGX desktops from Nvidia are beyond awesome. They come directly with a unified memory (a lot of it), and directly in Linux (stable, no need to waste time getting rid of Windows). They are not your usual consumer desktops (pricing starts at about $4000), but for any business/programmer focused on robotics, it may become the way to go, especially when considering the Omniverse ecosystem, which requires Nvidia hardware. #ai #robotics #omniverse

🚀 Classic vs. Adaptive AUTOSAR: Decoding Automotive’s Twin Pillars (No Fluff, Just Key Insights) The automotive software revolution runs on two engines: Classic AUTOSAR (CP) and Adaptive AUTOSAR (AP). Here’s how they differ and coexist → 🔹 Core Philosophies: Classic AUTOSAR → Your go-to for safety-critical, deterministic systems. Think airbags, engine control, or braking (ASIL-D). Runs on lean microcontrollers with hard real-time guarantees (μs deadlines). Adaptive AUTOSAR → Built for high-compute, dynamic workloads. Powers infotainment, ADAS, or OTA updates. Thrives on POSIX OS (Linux/QNX) with soft real-time (ms deadlines). 🔹 Critical Differences: → Programming: C (CP) vs. C++ (AP) → Communication: Static CAN/LIN signals (CP) vs. dynamic service-oriented networking (SOME/IP over Ethernet in AP) → Updates: Full ECU reflash (CP) vs. incremental OTA updates (AP) → Hardware: Resource-constrained MCUs (CP) vs. GPU-enabled SoCs (AP) 🔹 Why Both Matter: Classic AUTOSAR remains the unshakable foundation for safety (ASIL-D certified). Adaptive AUTOSAR enables software-defined innovation (AI, cloud connectivity, “automotive app stores”). They talk through smart gateways—converting CAN signals to Ethernet services for seamless collaboration. 🔹 When to Choose: • Classic AUTOSAR: If you need μs precision for brakes, powertrain, or safety systems. • Adaptive AUTOSAR: For AI sensor fusion, cockpit systems, or over-the-air features. • Hybrid: Domain controllers merging both (e.g., adaptive ADAS + classic vehicle control). 🚀 The Future: Adaptive AUTOSAR is growing at >20% CAGR as E/E architectures centralize, but Classic won’t fade—safety never compromises. The synergy enables true software-defined vehicles. Takeaway: CP = Certified Predictability | AP = Agile Potential. #AUTOSAR #Automotive #EmbeddedSystems #ADAS #SoftwareDefinedVehicle #AutoTech #CP #AdaptiveAUTOSAR #FutureMobility

🚗 What is AUTOSAR and why does it matter? In the past, automotive software was tightly linked to hardware. If an ECU or controller changed, the application had to be redeveloped from scratch. Suppliers also struggled to reuse the same code for different OEMs, leading to higher cost and complexity. AUTOSAR solves these issues. By separating hardware and software, it enables standardization, reusability, and scalability. The same software can now run on different platforms—from small ECUs to advanced ADAS systems. 📌 Two main pillars of AUTOSAR (Automotive Open System Architecture): 1️⃣ Classic AUTOSAR • Targets real-time, safety-critical functions (engine, braking). • Uses RTE (Runtime Environment) to connect SWCs and BSW (Basic Software). • Optimized for resource-constrained microcontrollers. 2️⃣ Adaptive AUTOSAR • Designed for high-performance ECUs in ADAS, autonomous driving, and connectivity. • Based on POSIX OS, supports service-oriented communication (SOA). • Enables dynamic updates (OTA) and runs on multicore CPUs. ⚙️ Why it matters • Standardizes collaboration between OEMs and suppliers. • Ensures compliance with ISO 26262 and safety requirements. • Accelerates the move towards software-defined vehicles (SDV). ⚙️ AUTOSAR Architecture • ASW (Application Software): Implements vehicle functions. • RTE (Runtime Environment): Middleware connecting ASW ↔ BSW. • BSW (Basic Software): Services, drivers, communication stacks, hardware abstraction. • Hardware: MCU, sensors, actuators. 🧩 Compositions, Connectors, Runnables & Events Applications are built from SWCs, grouped into Compositions (subsystems). Connectors link ports between SWCs/Compositions, using standardized interfaces. Inside each SWC, the smallest executable unit is a Runnable, triggered by Events (timing, data reception, or function calls). This makes AUTOSAR modular and event-driven. 🔄 Interfaces • Sender-Receiver: Asynchronous data exchange (sensor values, actuator commands). • Client-Server: Function calls, synchronous or asynchronous (diagnostics, services). 💾 Persistent Data (NVM SWC) When data must survive resets—like calibration values or fault counters—the NVM SWC defines it as NvM blocks and uses NvM services for storage in Flash/EEPROM. 👉 Think: NVM SWC = I want to save data → NvM BSW = I’ll store it safely. ✅ Takeaway: • RTE: Connects ASW ↔ BSW • SWCs & Runnables: Enable modular, event-driven design • Interfaces (Sender-Receiver / Client-Server): Manage data exchange and function calls • NVM SWC & NvM: Safely store persistent data ✨ Final thought: AUTOSAR is more than an ECU architecture. It’s the foundation of modular, event-driven, and service-oriented automotive software—enabling the shift to software-defined vehicles.   #AUTOSAR #AutomotiveSoftware #ADAS #SoftwareDefinedVehicle #EmbeddedSystems #Bertrandt

The next generation of gadgets will be... robots. We’re starting to see small AI-powered machines like this robot [MARS] that can pick up socks, explore your house, and even send an email if something’s wrong. At first, people will play with them the same way they downloaded apps when the App Store launched: experimenting, teaching them simple tricks, sharing “skills” with friends. The difference is that instead of just tapping a screen, you’ll be programming physical behavior. Your robot won’t just launch Spotify; it’ll fold laundry, check the oven, or play chess with you. You’ll be able to send it out to buy groceries or take care of your backyard. The possibilities are endless. And the easier these tools become to train and extend, the faster they’ll go from toys to household essentials. Pretty cool! Follow me The AI Colony for more. #ainews #breakingnews #aitools

Qualcomm, the giant chipmaker best known for powering smartphones, is making a major move into the robotics industry. The company announced on Tuesday that it

🚗 Android Auto Updates: Greater Driving Safety Hello! In the world of smart mobility, Google continues to innovate with Android Auto. The latest update introduces key features to reduce distractions while driving, such as intelligent call filtering and voice note-taking. These improvements, powered by AI, enable a safer and more productive experience on the road. Let's break down the essentials. 📞 Intelligent Call Filtering This new tool analyzes incoming calls and automatically categorizes them, silencing non-urgent ones so they don't interrupt your focus. You'll only receive alerts from important contacts or known numbers, minimizing risks on the road. ✅ Reduces distractions: Avoids unnecessary interruptions during driving. ✅ Customization: Set filters based on your preferences and usage patterns. ✅ Native integration: Works without additional apps, directly in the Android Auto interface. 🗣️ Real-Time Voice Note-Taking Now, you can dictate quick notes while driving, and the AI transcribes them automatically into text. Ideal for remembering ideas, tasks, or reminders without letting go of the wheel or looking at your phone. ✅ Accurate transcription: Uses advanced voice recognition to capture details precisely. ✅ Easy access: Notes are saved in Google Keep or compatible apps, ready to review upon arrival. ✅ Safety first: Everything is processed in the background, without requiring visual interaction. These updates arrive with the latest version of Android Auto, available for compatible devices. They represent a step forward in AI integration for mobile environments, prioritizing road safety in an increasingly connected ecosystem. For more information, visit: https://enigmasecurity.cl, check before completing the domain this email: enigmasecurity.cl #AndroidAuto #MobileTechnology #RoadSafety #ArtificialIntelligence #GoogleInnovation #SmartMobility Connect with me on LinkedIn to discuss more about tech and cybersecurity trends: https://lnkd.in/eFb3bY4C 📅 Tue, 30 Sep 2025 20:44:23 +0200 🔗Subscribe to the Membership: https://lnkd.in/eh_rNRyt