self-drive
To drive autonomously; (of a motor vehicle) to drive itself without a human operator, by means of sensors and computer technology.
self-drive: a vehicle that steers itself using sensors and code
Self-driving means a motor vehicle operates without human input on steering, throttle, or braking by using sensors, processors, and control algorithms. The vehicle perceives its environment through cameras, radar, lidar, or ultrasonic sensors, processes that data in real time, and executes driving decisions automatically. This is distinct from cruise control or lane-keeping assist, which handle only specific functions while a human driver remains in control and responsible.
Automation levels range from Level 2 (partial automation: steering or acceleration automated, but not both) to Level 5 (full automation under all conditions, with no steering wheel required). Most production vehicles with "self-driving" features operate at Level 2 or 3, meaning they can handle highway driving or parking maneuvers but require a human to monitor the road and intervene if systems fail. True Level 4 and 5 autonomy remains limited to controlled environments like robotaxis in defined geographic zones or autonomous shuttles at specific facilities.
How the systems work
A self-driving vehicle fuses data from multiple sensor types: cameras read road markings and traffic signals at high resolution; radar measures distance and velocity of nearby objects through rain or fog; lidar generates 3D point clouds of the scene by scanning with laser pulses. Onboard computers running machine learning models interpret this input, predict the behavior of other road users, and generate steering angles and acceleration or braking commands. Redundancy is critical: if one sensor or compute module fails, backup systems must still keep the vehicle safe.
Failures occur when sensors are obscured (heavy snow covering a lidar unit), when the environment is unfamiliar (a street the training data never included), or when edge cases overwhelm the decision logic. A self-driving system may freeze at an unmarked intersection or misinterpret a road construction pattern. Weather, glare, and deliberately adversarial objects (graffiti on a lane marker) can degrade performance. This is why deployment has been slower than early enthusiasts predicted, and why human responsibility and liability remain unclear in many jurisdictions.
The term "self-drive" entered common usage with Tesla's Autopilot (2015) and Waymo's public trials (2017), though it often inflates the actual capability of the systems in question. Regulators and insurers now scrutinize marketing language carefully, since calling a Level 2 system "self-driving" invites driver complacency and accidents. In practice, the term is most accurate when applied to vehicles operating within defined operational design domains: specific roads, specific speeds, specific weather conditions, where the automation has been thoroughly validated.