AV Glossary
Every autonomous driving term you need to know, from ADAS to World Models. Bookmark this page as your reference guide.
- ADAS (Advanced Driver Assistance Systems)
- An umbrella term for electronic systems in a vehicle that assist the driver. Includes features like adaptive cruise control, lane-keeping assist, automatic emergency braking, and blind-spot monitoring. ADAS covers SAE Levels 1 and 2.
- ADS (Automated Driving System)
- The hardware and software that can perform the entire dynamic driving task on a sustained basis. Used to describe Level 3-5 systems where the vehicle, not the human, is the driver during system operation.
- AEB (Automatic Emergency Braking)
- A safety feature that detects an imminent collision and applies the brakes automatically if the driver does not respond in time. Required on all new US vehicles starting in 2029 under NHTSA rules.
- AV STEP
- ADS-equipped Vehicle Safety, Transparency, and Evaluation Program. NHTSA's framework for national reporting, exemptions, and safety standards for autonomous vehicles, proposed in 2025 and being implemented in 2026.
- C-V2X (Cellular Vehicle-to-Everything)
- A V2X communication standard based on 3GPP LTE and 5G cellular technology. Offers greater range than DSRC by leveraging existing cell tower infrastructure. Increasingly favored over DSRC for next-generation connected vehicle deployments.
- Cybercab
- Tesla's purpose-built robotaxi vehicle with no steering wheel or pedals, designed exclusively for autonomous ride-hailing. First unit produced in February 2026 at Gigafactory Texas.
- Deadheading
- When a robotaxi drives empty between passenger fares (returning to a pickup zone or repositioning). Deadheading miles add to congestion without serving passengers and are a key metric for robotaxi fleet efficiency.
- Disengagement
- An event where the autonomous driving system deactivates and hands control back to a human driver or remote operator, either because it encounters a situation it cannot handle or the human manually takes over.
- DMS (Driver Monitoring System)
- An interior camera system that tracks the driver's eye gaze, head position, and alertness to determine if they are paying attention. Required in all new EU vehicles by July 2026. Critical for Level 2+ and Level 3 vehicles.
- DSRC (Dedicated Short-Range Communication)
- A WiFi-based radio communication standard using the 5.9 GHz band, designed for vehicle-to-vehicle and vehicle-to-infrastructure communication. A mature technology being gradually supplanted by C-V2X.
- Dynamic Driving Task
- All the real-time operational and tactical functions required to operate a vehicle in on-road traffic. Includes steering, braking, accelerating, monitoring the environment, and responding to events.
- End-to-End Neural Network
- An autonomous driving architecture where a single neural network processes raw sensor inputs and directly outputs driving decisions (steering, acceleration, braking), without separate hand-coded modules for perception, prediction, and planning.
- FMVSS (Federal Motor Vehicle Safety Standards)
- US federal regulations prescribing minimum safety performance for motor vehicles. Originally designed for human-driven vehicles, NHTSA is updating several FMVSS standards to accommodate driverless vehicles.
- FSD (Full Self-Driving)
- Tesla's branded driver assistance software package. Despite the name, FSD is currently classified as SAE Level 2, meaning the human driver must supervise the system at all times.
- Geofencing
- Defining a virtual geographic boundary within which an autonomous vehicle is authorized to operate. Most Level 4 robotaxis operate within specific geofenced areas that have been mapped and validated.
- HD Map (High-Definition Map)
- A centimeter-accurate 3D map of the road environment including lane markings, curbs, traffic signs, and fixed obstacles. Used by companies like Waymo as a reference layer for autonomous navigation. Some newer systems (Tesla, XPeng) aim to operate without HD maps.
- LiDAR (Light Detection and Ranging)
- A sensor that fires laser pulses and measures their reflections to build a precise 3D point cloud of the surrounding environment. Provides exact distance measurements to every detected object. Used by Waymo, Zoox, and most commercial robotaxi operators.
- Minimal Risk Condition (MRC)
- A safe state that an autonomous vehicle achieves when it can no longer operate normally. Typically involves a controlled deceleration, pulling to the road shoulder, activating hazard lights, and notifying a remote operator.
- ODD (Operational Design Domain)
- The specific conditions under which an autonomous driving system is designed to operate: geographic area, road types, speed range, weather conditions, and time of day. A Level 4 system must achieve full autonomy within its ODD.
- OTA (Over-the-Air Update)
- Software updates delivered wirelessly to a vehicle, similar to smartphone updates. Allows manufacturers to improve autonomous driving performance, fix bugs, and add features without requiring a physical service visit.
- Phantom Braking
- When an ADAS or automatic emergency braking system applies the brakes without cause, typically because the perception system falsely detects an obstacle. Associated with camera-only systems misinterpreting shadows, overpasses, or reflective surfaces.
- Point Cloud
- A 3D dataset generated by LiDAR consisting of millions of points, each with an x/y/z coordinate. Provides a detailed spatial representation of the vehicle's surroundings, including objects, road surfaces, and structures.
- Robotaxi
- An autonomous vehicle operating as a ride-hailing taxi without a human driver. Current operators include Waymo (US), Baidu Apollo Go (China/Dubai), Zoox (US), and Tesla (US, with safety monitor).
- SAE Levels (0-5)
- The Society of Automotive Engineers' six-level classification of driving automation, from Level 0 (no automation) to Level 5 (full automation in all conditions). The global standard for describing autonomous driving capability.
- Sensor Fusion
- The process of combining data from multiple sensor types (cameras, LiDAR, radar, ultrasonic) to create a unified, more accurate perception of the environment. Waymo's multi-sensor approach is the leading example.
- Teleoperations
- Remote human monitoring and control of an autonomous vehicle. A teleoperator can guide a vehicle through a situation the autonomous system cannot handle, such as a blocked road or unusual construction zone.
- TOPS (Trillion Operations Per Second)
- A measure of computing performance used to describe the processing power of autonomous driving chips. XPeng's Turing chip delivers 3,000 TOPS per vehicle; higher TOPS enables more complex neural network processing.
- V2X (Vehicle-to-Everything)
- Communication technology enabling vehicles to exchange data with other vehicles (V2V), infrastructure like traffic lights (V2I), pedestrians (V2P), and cloud networks (V2N). NHTSA estimates V2X could save 1,000 lives per year.
- Vision-Only
- An autonomous driving approach that relies exclusively on cameras and neural networks for perception, without LiDAR or radar. Championed by Tesla. Lower hardware cost but more vulnerable to poor visibility conditions.
- VLA (Visual-Language-Action Model)
- A next-generation AI architecture that combines vision, language understanding, and driving action in a single model. Integrates large language models to give autonomous vehicles common-sense reasoning and the ability to interpret text, signs, and verbal commands.
- World Model
- An AI model that maintains an internal simulation of how the physical world works, predicting what will happen next based on current observations. Used in next-generation autonomous driving to anticipate the behavior of other road users.
Missing a term?
The autonomous driving field evolves rapidly. If you cannot find a term, let us know and we will add it.
Suggest a Term