Robots Atlas>ROBOTS ATLAS
Standard

Level 4 (L4)

2014ActivePublished: 24 October 2026Updated: 24 October 2026Published
Key innovation
Under SAE J3016, Level 4 shifts the dynamic-driving-task fallback from the human to the system itself: within a defined operational design domain the vehicle needs no driver ready to intervene.
Category
Standard
Abstraction level
Standard
Operation level
SystemRobot control
Use cases
Robotaxi / autonomous ride-hailingAutonomous shuttles and busesGeofenced mobility servicesAutonomous last-mile deliveryClosed-campus and industrial-zone transportHighway autopilot within the ODD

How it works

The ADS integrates perception (camera, LiDAR and radar sensor fusion), localization (HD maps, GNSS/IMU), prediction and trajectory planning, and control modules. It operates only within a defined operational design domain (ODD) and continuously monitors whether conditions remain inside that domain. When a situation exceeds the system's capability or the ODD boundary, the ADS executes a fallback maneuver to a minimal risk condition on its own (e.g. pulling over and stopping), without relying on a human to take over.

Problem solved

It addresses the central weakness of Level 3 (Conditional Automation), where the system drives but the human remains the fallback and must be ready to respond to a request to intervene. At Level 4 the system itself handles the fallback, which within the ODD removes the need for human supervision and enables driverless operation such as robotaxis.

Components

Perception moduleEnvironment perception

Fuses data from cameras, LiDAR, radar and sonar to detect vehicles, pedestrians and obstacles.

Localization moduleVehicle positioning

Combines HD maps / 3D point clouds, GNSS and IMU to pinpoint the vehicle's position and orientation.

Planning & prediction moduleMotion decision-making

Predicts the behavior of other road users and computes a safe trajectory (speed, steering angle).

Control / actuation moduleManeuver execution

Translates the planned trajectory into acceleration, braking and steering commands sent to the vehicle.

ODD monitoring & fallback (MRC)Fallback handling

The component that distinguishes L4: monitors ODD boundaries and, when they are exceeded or the system fails, brings the vehicle to a minimal risk condition on its own without driver involvement.

Evolution

Original paper · 2014 · SAE International · SAE On-Road Automated Driving (ORAD) Committee
Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles (SAE J3016)
SAE On-Road Automated Driving (ORAD) Committee
2014
SAE J3016 published
Inflection point

SAE International publishes the standard defining six levels of driving automation (0–5), including Level 4.

2016
J3016_201609 revision and NHTSA adoption

Updated revision of the standard; the US NHTSA adopts the SAE level classification.

2020
First commercial driverless robotaxi service
Inflection point

Waymo becomes the first to offer driverless rides within a limited operational design domain (Phoenix), demonstrating L4 in practice.

2021
Current revision J3016_202104

Latest revision of the standard, refining ODD, fallback and minimal-risk-condition terminology.

2023
L4 legal framework in Japan

Japan's amended Road Traffic Act introduces an L4 protocol; ZEN drive Pilot becomes the first approved L4 device.

2024
Driverless L4 robotaxi in the US

Waymo runs a commercial robotaxi service in parts of the US entirely without safety drivers.

Hyperparameters (configurable axes)

Operational Design Domain (ODD)Critical

The set of conditions under which an L4 system operates autonomously: geographic area, speed range, time of day, weather, road types. It defines scope and is key to distinguishing L4 (bounded ODD) from L5 (no ODD limits).