PatentBrief

Patent Landscape

Patent Landscape:
Autonomous Vehicles

Waymo sued Uber for $245 million over stolen self-driving IP. The autonomous vehicle industry is built on patents — and the safety-critical nature of the technology makes IP disputes uniquely consequential.

The autonomous vehicle industry has produced one of the most intensive patent races in automotive history. Waymo alone holds over 1,500 autonomous vehicle patents, and the field is defined by the strategic tension between sensor-heavy approaches (LIDAR-based mapping, pioneered by Waymo) and vision-only approaches (camera and neural network-based, championed by Tesla). These technical choices are reflected in patent portfolios that are largely incompatible — making cross-licensing difficult and IP disputes likely as the industry consolidates.

The AV patent landscape divides into perception (how the vehicle sees the world), prediction (how it anticipates other actors), planning (how it decides what to do), and mapping (what prior knowledge it relies on). Each layer has its own IP dynamics and dominant players. The companies that control the most valuable patents in each layer will determine the economics of autonomous vehicle licensing for the next two decades.

Key Patents

US9,836,0522016

Autonomous Vehicle Sensor Fusion and 3D Environment Mapping

Waymo (Google)

The central patent in Waymo's autonomous vehicle portfolio — covering the method of fusing LIDAR, radar, and camera data into a real-time 3D world model. This patent was at the heart of Waymo's landmark lawsuit against Uber, which settled for $245 million when Uber's self-driving team was found to have used stolen Waymo trade secrets and IP.

US10,289,1142019

Autopilot System with Vision-Based Lane Keeping and Traffic Awareness

Tesla Motors

Tesla's Autopilot patent covers the vision-only approach to autonomous driving — using neural networks trained on fleet data to detect lanes, traffic signs, and other vehicles without LIDAR. Tesla's camera-first strategy is a deliberate technical and patent differentiation from Waymo's LIDAR-heavy approach.

US10,795,3712020

HD Map Generation and Localization for Autonomous Vehicles

Mobileye (Intel)

Mobileye's HD mapping patent covers the method of crowdsourcing map data from camera-equipped vehicles to build and continuously update centimeter-accurate maps for autonomous navigation. Mobileye's REM (Road Experience Management) system has mapped over 1 billion kilometers of road — this map data IP is as valuable as the sensor technology itself.

US11,048,2562021

Behavior Prediction for Vulnerable Road Users Using Recurrent Networks

Argo AI (Ford / VW)

Predicting the future movements of pedestrians, cyclists, and other unpredictable road users is the hardest problem in autonomous driving. This patent covers the LSTM-based prediction system for modeling the intent and trajectory of vulnerable road users — a critical safety requirement for urban autonomous vehicle deployment.

US10,627,8252020

Vehicle-to-Everything (V2X) Communication for Intersection Negotiation

Qualcomm

Qualcomm's V2X patent covers the protocol by which autonomous vehicles communicate with each other and with roadside infrastructure to negotiate priority at intersections without traditional traffic signals. V2X is the cooperative layer that enables autonomous vehicles to operate beyond what any single vehicle's sensors can perceive.

US10,726,2782020

Reinforcement Learning for Autonomous Driving Policy in Complex Scenarios

NVIDIA

NVIDIA's Drive platform uses reinforcement learning to train autonomous driving policies in simulation before deployment. This patent covers the method of generating synthetic training scenarios and using simulation-learned policies on real vehicles — the approach that reduces the billions of real-world miles required to validate safety-critical autonomous systems.

Key Players

Waymo

The most valuable and most defensively patent-protected autonomous vehicle program. Waymo's portfolio covers every layer of its stack — sensor design, perception algorithms, behavior prediction, and routing. The Uber lawsuit established Waymo's willingness to litigate aggressively. As the only commercial robotaxi operation at scale (in Phoenix and San Francisco), Waymo's patents are actively defining what operational AV IP looks like.

Tesla

Tesla's Full Self-Driving program generates a fleet learning advantage that is partially captured in patents but largely protected as trade secrets — the neural net weights trained on millions of Tesla vehicles cannot be easily reverse-engineered. Tesla's patent strategy focuses on the hardware (FSD chip architecture) and specific software methods, while the trained models themselves remain trade secrets.

Mobileye

Intel-acquired Mobileye supplies AV perception systems to BMW, Volkswagen, GM, and dozens of other automakers. Mobileye's IP strategy is dual: ADAS sensor IP for near-term revenue, and HD mapping plus REM crowdsourcing IP for long-term autonomous platform control. The company that owns the maps controls a critical autonomous vehicle input.

Qualcomm

Positions itself as the connectivity and compute platform for autonomous vehicles — not building full AV stacks but owning the V2X communication patents and the Snapdragon automotive chip IP that other AV programs run on. Qualcomm's horizontal platform strategy mirrors its approach in mobile: own the essential infrastructure patents that every vertical player must license.

What to Watch

01

Simulation-to-Reality Transfer Patents

Training autonomous vehicle systems in simulation and deploying them in the real world requires solving the domain gap between synthetic and real sensor data. The patents covering synthetic data generation, domain randomization, and simulation validation methods are becoming critical IP as regulators require extensive testing before deployment — and simulation is the only cost-effective path to sufficient test coverage.

02

AV Safety and Liability IP Framework

As autonomous vehicles enter commercial deployment, the patents covering how safety decisions are made — when to engage human override, how to handle sensor failures, what to do in unavoidable collision scenarios — are taking on regulatory and legal significance. The decision-making IP may directly determine legal liability when accidents occur.

03

Autonomous Trucking Corridor Patents

Aurora Innovation, Kodiak Robotics, and Torc Robotics are focusing autonomous vehicle IP on long-haul trucking on defined highway corridors — a significantly simpler technical problem than urban robotaxis with clearer near-term commercialization. The AV trucking companies are building narrow, defensible patent portfolios around the specific operational conditions of highway freight.

From PatentBrief

Explore AV patents on PatentBrief →

Search autonomous driving, sensor fusion, and vehicle navigation patents. Read any patent in plain English and understand the claims that define the self-driving future.

Search AV patentsAll patent landscapes →