Self-driving car

A road vehicle that uses sensors and software to perform some or all driving tasks without continuous human control. Its automation may still require a human to supervise or take over.

Connect

Automotive perception: The process of interpreting sensor data to identify road features, objects, and traffic conditions. Perception turns camera, radar, and lidar measurements into a usable picture of the road.
Automated driving system: The hardware and software that perform the dynamic driving task in a defined operating context. This is the component responsible for driving when automation is engaged.
Robotaxi: A passenger vehicle operated as an on-demand taxi using automated driving technology. Robotaxis are a visible application of driverless operation in defined service areas.
Human driving: The operation of a vehicle through a human driver's perception, judgment, and control. Human drivers remain the reference point for responsibility and performance comparisons.
Automated vehicle safety: The assessment and management of risks associated with vehicles that perform driving tasks automatically. Safety claims require evidence across rare events as well as ordinary driving.
Sensor fusion: The combination of measurements from multiple sensors to produce a more reliable estimate of the world. Combining complementary sensors helps compensate for the limits of any single sensor.
SAE J3016: The SAE standard that classifies driving automation from Level 0 through Level 5. Its levels distinguish limited assistance from systems that can drive without human fallback.
Autonomous trucking: The use of automated driving systems to operate heavy trucks with reduced or absent human control. Long-haul and freight operations offer a distinct setting for vehicle automation.
Teleoperation: The remote control of a machine by a human using communications and control interfaces. Remote driving can assist vehicles without making them independently autonomous.
Trolley problem: A philosophical thought experiment about choosing between harmful outcomes when action is unavoidable. It frames public debate about how automated systems should handle unavoidable crashes.