Chapter 6: IoT - Autonomous Cars Flashcards
What does the Society of Automotive Engineers (SAE) do?
- Formally defined six different levels of automation
- Unified vocabulary for common understanding
- Helps to keep track of driverless technology advancement
What are the six level of automation and on which threshold do machines monitor the environment
0: No Automation
1: Driver Assistance
2: Partial Automation
- —- Automated Driving System Monitors Driving Environment (Machine takes over the majority of the tasks) ———
3: Conditional Automation
4: High Automation
5: Full Automation
According to the SAE levels, which tasks are either executed by a human driver or the system?
- Execution of steering and acceleration / deceleration
- Monitoring of driving environment
- Fallback performance of dynamic driving task
- System capability (driving modes)
Explain the Waymo - Formerly Googles Driverless Car
- Passed more than 5 million miles in tests
- Integrates driverless technology in existing cars
Vision: Fully autonomous - no steering wheel / pedals
Technology: sensors, cameras, embedded HW, LIDAR system
Explain the LiDAR technology
- Similar to radar, uses light pulses (millions per second)
- Distance to objects is measured when laser pulses bounce back to scanner attached on car (accuracy of +- 2cm)
Result: Accurate 3D Maps of the car environment
What is the formula for measuring distance with the LiDAR Technology?
d = (Et x c) / 2
d = distance Et = elapsed time c = speed of light
What are issues of LiDAR Technology?
- Dirty sensors
- Stabilization issues due to holes
- Dynamic traffic like police lights that could interfere with the light sensor
Explain the Mercedes Intelligent Drive “Bertha”
- Vision: Assisted driving
- Cameras for positioning and obstacle / traffic detection
- LiDAR for obstacle detection integrated into chassis
- More sensors -> more protection
What are the functions of mercedes intelligent drive?
- Front cameras: traffic sign recognition
- Back short range radar: for rear collision warning
- Front long-range radar: for adaptive cruise control
What are some issues with Mercedes intelligent drive?
- Weather impact on sensors
- Traffic light recognition at intersections
Explain the Tesla Autopilot and what is the challenge with it?
- Teslas strategy: Full autonomy using only cameras, radar, and ultrasonic sensors (no LiDAR)
- Still in beta version, keeps lane & speed but does not change the street at intersections
Challenge:
Accidents have been a source of controversy! (White truck could not be identified on a very sunny sky)
Describe the Autonomous Driving Architecture
The architecture runs in a control loop (Sense, Think, Act) that feeds the system.
Perception Layer: Software is linked to actuators / to the outside (camera / sensors) to build a model of the environment)
Interpretation and Planning: Understand the scene, predict what is going to happen in the scene, plan maneuvers
Control: Plan the trajectory and dynamically control the planned trajectory
What are other essential parts of autonomous cars except for driving?
- There are many embedded systems in a car
- Interconnected embedded systems work together in the car (media and control systems)
Which embedded systems exist in a car?
- Controller area network (CAN bus): designed for communication between microcontrollers and devices
- Electronic control unit (ECU): generic term for embedded system that controls electrical systems in a vehicle (brake control module, Engine control module)
ECUs are often connected through the CAN
What are the security implications of autonomous driving?
- Its possible to hack a car and to control almost all functionalities
- Hack was possible because CAN was not separated from entertainment system
- After entering CAN almost every ECU could be controlled
