Innovations in LIDAR with Low-Cost Components

A LiDAR system for autonomous vehicles uses two sensors, one to measure the peak reflection intensity and another with a configurable trigger threshold to measure the time when the peak occurred. This allows accurate reflection intensity and time detection without an expensive high-speed ADC. The LiDAR uses a laser and sensors to measure the reflection peak from objects around the vehicle.

Generating high-resolution 3D point clouds from low-resolution LiDAR point clouds and camera images to enable autonomous vehicles to operate with lower-cost sensors. The system uses machine learning to combine a downsampled LiDAR point cloud with camera images to generate a higher-resolution depth map.

A LiDAR system that uses a pulsed time-of-flight (TOF) laser range-finding technique for measuring distances to objects. The LiDAR system has features to optimize cost while maintaining long-range, high resolution, and reliability. The features include collimated laser beams, averaging of multiple laser pulses, and motion synchronization to avoid temporal averaging. The LiDAR system uses arrays of solid-state lasers and detectors for compactness, reliability, and low cost.

LiDAR systems for autonomous vehicles use multiple wavelengths of light to improve performance and reduce size, cost, and complexity compared to single-wavelength LiDARs. The multi-wavelength LiDARs use lasers with different colors to enable higher-resolution 3D mapping without moving parts. The LiDAR sensors have modular designs that allow flexibility for different vehicle types. The sensors also have optical monitoring to ensure the lasers are operating correctly and detect interference.

LiDAR system with a 2D array of VCSEL lasers that can be individually addressed for LiDAR applications like autonomous vehicles. The LiDAR system simultaneously controls the lasers in the array to emit independent laser pulses to avoid crosstalk while minimizing the number of electrical drivers needed. The laser array is driven in a matrix-addressable manner by row/column to individually energize each VCSEL with a low voltage below its breakdown voltage. This allows 2D control of the laser array with a limited number of drivers.

A compact, low-cost rotating LiDAR system for autonomous vehicles that provides high positional accuracy, long-distance range, and low power consumption. The system has a stationary base with a rotating upper assembly that contains the LiDAR sensor. The rotation is achieved using a minimal contact, brushless electromagnetic motor. The rotating assembly communicates with the base wirelessly using optical and RF links. This eliminates the need for slip rings and allows compact integration. The wireless rotation and communication enable a simpler, lower-cost design compared to traditional rotating LiDAR systems.

LiDAR system with non-uniform vertical distribution of laser beams. The LiDAR rotor has multiple lasers that emit beams at different vertical angles. This concentrates the beams near the horizontal for better ground detection while using fewer lines. An optical splitter and rotating mirror scan the beams. The non-uniform distribution of vertically angled lasers helps optimize beam density for ground-level targets, like pedestrians and vehicles while reducing unnecessary vertical coverage.

Generating high-density 3D maps of a scene using low-density distance sensors and high-density image sensors. By fusing passive and active measurements from cameras, LiDAR, radar, etc., the method analyzes the 3D map and images to identify stationary vs moving objects. It then uses active distance measurements only on moving objects to estimate distances to stationary objects. This allows the generation of high-density 3D maps from low-density sensors, improving resolution, coverage, and cost over using just LiDAR alone.

Localization for autonomous vehicles using monocular cameras instead of expensive LiDAR scanners. The system compares camera images to a 3D map to find the vehicle's location. Synthetic camera views are generated and matched to maximize mutual information. The camera-based localization achieves similar accuracy to LiDAR-based methods at a much lower cost.

LiDAR sensor for autonomous vehicles that uses multiple low-cost VCSEL laser arrays and detectors instead of a single laser/mirror assembly. The VCSEL arrays are directed by lenses to cover the desired field of view. The laser beams are scanned by turning on/off individual lasers in each array. This creates a 3D point cloud of reflections from the detectors.