Autonomous Vehicle Patents

LiDAR (Light Detection and Ranging) is the most actively litigated technology area in autonomous vehicles, with major patent disputes having already reshaped the competitive landscape: VELODYNE — THE LIDAR ORIGINATOR: David Hall (Velodyne founder) invented the spinning 64-beam LiDAR (HDL-64E) that made autonomous vehicle research practical; Velodyne's core patents cover: the rotating mirror assembly; multi-beam spinning design; real-time point cloud generation; these patents were highly valuable through approximately 2020-2025 but are now expiring or have expired; Velodyne sued multiple competitors including Waymo and Hesai (Chinese lidar); WAYMO — CUSTOM LIDAR INNOVATION: Waymo developed its own custom LiDAR in-house to reduce costs (from ~$75,000 per unit to under $7,500 by 2017); key Waymo LiDAR patents: US9,368,936 (vertical array design); multiple patents on custom ASIC for LiDAR signal processing; solid-state flash LiDAR components; patents on integrating short-, mid-, and long-range LiDAR into a single perception stack; THE WAYMO VS. UBER LITIGATION: Anthony Levandowski, a senior Waymo engineer, downloaded 14,000 confidential Waymo files before leaving to start Otto, which was immediately acquired by Uber; Waymo sued Uber in 2017 for trade secret misappropriation and patent infringement; Settled in 2018: Uber paid ~$245 million in equity to Waymo; this case accelerated LiDAR IP awareness across the entire AV industry; SOLID-STATE LIDAR — NEXT GENERATION: traditional rotating LiDAR has moving parts that limit reliability; solid-state alternatives: MEMS-BASED: Innoviz Technologies (InnovizOne; InnovizTwo): BMW design win; Cepton Technologies (Honda partnership; acquired by Continental); flash LiDAR: FMCW (Frequency-Modulated Continuous Wave): Aeva: simultaneous range + velocity measurement from single pulse; Scantinel Photonics; Aurora Innovation; advantages: can detect velocity directly; immune to ambient light interference; CHINESE LIDAR MANUFACTURERS: Hesai Technology (US IPO 2023): one of the largest lidar suppliers by volume; RoboSense: strong Chinese market; DJI's Livox division; AUTOMOTIVE-GRADE REQUIREMENTS: most current LiDAR is not yet automotive grade (AEC-Q100; ISO 26262 ASIL-B/C/D); Luminar claims automotive-grade Iris LiDAR with 250m range at 10% reflectivity; achieving automotive grade is itself a patent moat.

The major autonomous vehicle companies have taken distinctly different technological approaches, each with different patent implications: WAYMO — SENSOR-RICH FULL AUTONOMY: technology approach: five types of sensors (radar; LiDAR; cameras; ultrasonic; custom compute); full L4 autonomy from the beginning; HD maps essential; patent strategy: deep patent portfolio in LiDAR design; HD mapping techniques; perception pipeline software patents; motion planning; occupancy grids; key differentiators: years of real-world Waymo One robotaxi data (>20 million autonomous miles by 2024); custom silicon (Waymo Researcher chip for LiDAR signal processing); TESLA — CAMERA-ONLY APPROACH: technology approach: camera-only (no LiDAR) for Full Self-Driving (FSD); occupancy networks replace point clouds; large language model influences on decision-making; Dojo custom AI training supercomputer; patent strategy: limited external publication; Tesla's real IP differentiation is data + compute scale (millions of vehicles collecting data 24/7); patents on: specific neural network architectures for camera-based 3D reconstruction; Dojo training cluster design; 4D occupancy networks; CONTROVERSY: Tesla's FSD has been criticized for L2 vs. L4 safety claims; Tesla calls it FSD but NHTSA classifies it as L2; GM CRUISE — MODERATE SENSOR SUITE: technology approach: LiDAR + radar + cameras; designed for specific operational domains; Cruise Origin vehicle designed specifically for robotaxi use (no steering wheel; no pedals); patent strategy: focused on dense urban environments; specific street-level mapping; difficult weather operation; Cruise suspended operations in 2023 after accident and resumed scaled-down operations in 2024; AURORA — ACQUISITIONS-DRIVEN: acquired LIDAR (OMAS); mapping (HERE HD map assets); radar capabilities; patents acquired through acquisitions add to organic portfolio; focus on trucking first (Aurora Horizon); MOBILEYE (INTEL ACQUISITION $15.3B; THEN IPO 2022): EyeQ chips: camera-based ADAS ASICs with massive installed base (hundreds of millions of vehicles); REM (Road Experience Management): crowd-sourced HD mapping from production vehicles; RSS (Responsibility Sensitive Safety): formal mathematical model for AV safety decisions; SuperVision: camera-based advanced driver assistance aiming toward autonomy; unique position: both a supplier AND developing its own AV platform.

Beyond perception hardware, the autonomous vehicle software stack — including motion planning, trajectory prediction, and decision-making — is a rich source of valuable patents: MOTION PLANNING: the problem: given a 3D perception of the environment, plan a safe trajectory for the next few seconds; APPROACHES: SAMPLING-BASED PLANNING: randomly samples possible trajectories and selects best; Rapidly-exploring Random Trees (RRT); Probabilistic Roadmaps; OPTIMIZATION-BASED: formulates trajectory planning as constrained optimization; Model Predictive Control (MPC); widely patented at academic spinoffs and OEMs; LATTICE-BASED: precomputed trajectory library on a grid; OCCUPANCY GRIDS AND MAPS: Carnegie Mellon spin-off techniques: occupancy grids for environment representation; many Waymo patents build on CMU foundations; RESPONSIBILITY SENSITIVE SAFETY (RSS): Mobileye's formal mathematical model for safe AV behavior; defines safety corridors and proper responses; US patents on the specific RSS formulas and their implementation; PREDICTION — WHERE OTHER AGENTS ARE GOING: trajectory prediction for pedestrians, cyclists, vehicles: social force models (early academic); neural network-based prediction; attention-based transformer models (most current); interaction-aware prediction; NVIDIA and Mobileye both hold significant prediction patents; HD MAPPING AND LOCALIZATION: HD maps are essential for most L4 AV approaches (Waymo; Cruise; Aurora); TOMTOM HD MAP patents; HERE HD Live Map (owned by Audi/BMW/Daimler consortium + Intel); SIMULTANEOUS LOCALIZATION AND MAPPING (SLAM): when operating without HD maps; Bosch; Continental; academic spinoffs; MOBILEYE REM: each production Mobileye-equipped car contributes anonymized mapping data; resulting crowd-sourced map covers 1 billion km; V2X COMMUNICATION: DSRC (Dedicated Short-Range Communications): 5.9 GHz band; Qualcomm; Bosch; Continental; C-V2X (Cellular V2X): LTE-V2X and 5G NR-V2X; Qualcomm key patent holder; V2X enables vehicles to communicate with each other and infrastructure.

The autonomous vehicle patent strategy differs significantly depending on whether you are a startup, a Tier 1 supplier, or an OEM: AV STARTUP PATENT STRATEGY: FOCUS ON DEFENSIBILITY, NOT OFFENSE: AV startups typically cannot out-patent the major players; defensive patenting protects from copycat behavior and provides cross-license leverage; FILE ON TECHNICAL INNOVATIONS, NOT SYSTEM ARCHITECTURE: claim the specific algorithm; the specific sensor calibration method; the specific data structure — not the broad concept of autonomous driving; TRADE SECRET VS. PATENT: given the rapid evolution of the space, trade secret protection for core algorithms is often as valuable as patents; Tesla's approach (minimal external disclosure; data moat) is the extreme version; if disclosure is unavoidable (academic publication; conference), file patents before disclosing; SENSOR DATA PATENTS: specific compression methods for LiDAR point cloud data; specific sensor fusion algorithms that combine camera + radar + LiDAR; these are technically specific and defensible; ACQUISITION TARGET VALUE: a startup with well-drafted, technically specific patents is more valuable as an acquisition target; acquirers pay premium for clean IP with strong priority dates; TIER 1 SUPPLIER STRATEGY: Tier 1 suppliers (Continental; Bosch; Aptiv; ZF Friedrichshafen; Denso) need both offensive and defensive portfolios; OFFENSIVE: patents on specific ADAS/AV component innovations; DEFENSIVE: broad patent portfolio to enable cross-licensing with OEMs and other Tier 1s; STANDARD ESSENTIAL: V2X, safety, and communication standards create SEP considerations; Tier 1s participating in standards bodies must manage SEP disclosure; OEM STRATEGY: OEMs are filing increasingly in software and AI areas (traditionally dominated by software companies); TOYOTA: largest automotive patent filer globally; strong in hydrogen AND AV; FORD: filed heavily in AV software patents post-2015; GM Cruise: strong combined OEM + AV subsidiary portfolio; KEY REGULATORY CONSIDERATIONS: NHTSA AV safety reporting requirements can reveal IP; FDA comparison: unlike medical devices, AV software has no clear approval pathway yet; ISO 26262 (functional safety) and SOTIF (ISO 21448) compliance creates design constraints that influence patentable implementations.