La voiture autonome : comment ça marche ? - L'Esprit Sorcier
Introduction to Autonomous Vehicles
The Race for True Autonomy
- The global automotive industry is in a race to develop fully autonomous vehicles capable of navigating real traffic without human intervention.
- There are various levels of vehicle autonomy, ranging from Level 0 (no automation) to Level 5 (full automation).
Levels of Vehicle Autonomy
Level 0: No Automation
- At Level 0, the driver controls all driving parameters and must constantly monitor the environment, similar to traditional cars without electronic assistance.
Level 1: Driver Assistance
- In Level 1, drivers can delegate specific tasks like acceleration or braking but must remain vigilant at all times.
Level 2: Partial Automation
- Vehicles at this level can manage both speed and steering simultaneously, such as automatic parking systems; however, drivers must still supervise.
Level 3: Conditional Automation
- At Level 3, under certain conditions (e.g., traffic jams), the car can take over monitoring while allowing the driver brief distractions.
Level 4: High Automation
- In this stage, vehicles operate independently on designated roads without driver oversight but are limited to specific environments like highways.
Level 5: Full Automation
- Fully autonomous vehicles (Level 5) can navigate any road and weather condition without human input or even a steering wheel.
The Technology Behind Autonomous Vehicles
Sensors and Data Collection
- To achieve autonomy, cars require advanced sensors including cameras, radars, LIDAR systems that gather essential data about their surroundings.
Functionality of Sensors
- Cameras help maintain lane position by detecting road markings and reading traffic signs. They also identify other vehicles and pedestrians.
Redundancy in Information Gathering
Understanding Autonomous Vehicle Navigation
The Role of 3D Maps in Localization
- With established 3D road maps, the vehicle can localize itself with centimeter precision, allowing it to follow routes accurately and anticipate upcoming turns or traffic light positions.
- This high level of accuracy instills confidence in passengers regarding reaching their destination safely.
Decision-Making in Autonomous Vehicles
The Brain of the Vehicle
- All raw data collected from various sensors is sent to the car's "brain," which is its software system. This software acts as the pilot by analyzing and interpreting data in real-time.
- The system employs artificial intelligence to make sense of sensor inputs, such as those from cameras and LIDAR, enabling it to detect objects around it.
Interpreting Human Presence
- The software can recognize human shapes and interpret them as pedestrians. This capability allows the vehicle to predict pedestrian behavior, such as entering a crosswalk.
- Based on this interpretation, the software decides to activate brakes when necessary for safety.
Questions About Decision-Making Authority