LiDAR Detector Array Volumetric Analysis for Object Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current LiDAR systems face limitations in high throughput and high resolution necessary for autonomous vehicle navigation, with inadequate 3D point cloud approaches for real-time object identification and feature extraction, especially in applications like autonomous vehicle navigation and mobile devices.
Innovation Solution
LiDAR systems utilize one or more emitters and a detector array to sample incoming signal intensity at a predetermined frequency, generating multiple samples per light packet for volumetric analysis, with a GPU interpreting the retroreflected signal to produce multiple output points, differentiate between objects, and determine attributes like slope and edge features, using frame buffers to account for environmental factors and calibration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If flash LiDAR illuminates entire 2D field of view with blanket of light, then simultaneous measurement of return values is achieved, but incident laser power available for each location becomes insufficient
Solution Approach 1:
The patent divides the field of view into multiple angular directions and uses multiple emitters to illuminate different portions of the scene simultaneously. Each emitter is assigned a specific angular sector, allowing the system to maintain high throughput while ensuring sufficient laser power is directed at each location. This segmentation of the illumination task resolves the contradiction between measuring the entire field of view quickly and providing adequate power to each point.
Solution Approach 2:
The patent introduces a temporal dimension by sampling the returned light packets at multiple predetermined frequencies during the emitter cycle. This allows the system to gather volumetric analysis data from the same spatial illumination, effectively increasing the information yield without requiring additional laser power. The multi-frequency sampling in the time domain compensates for the reduced spatial power density.
2Productivity
If multiple light packets are emitted in rapid succession, then high measurement throughput is achieved, but resolution and accuracy of object identification deteriorates
Solution Approach 1:
The patent performs preliminary volumetric analysis of the light packet structure by sampling at multiple frequencies before final object identification. This preliminary processing extracts key features and characterizes the return signal in advance, allowing the system to maintain high throughput while preserving identification accuracy. The preliminary volumetric characterization prepares the data for rapid, accurate object recognition without requiring slower, more precise measurements.
Solution Approach 2:
The system uses feedback from the volumetric analysis of returned light packets to refine object identification. By analyzing the multi-frequency sampled data, the system can identify objects more accurately and use this information to optimize subsequent measurements. This feedback loop allows high throughput to be maintained while improving, rather than sacrificing, identification accuracy.
3Area of stationary object
If 3D point cloud approach is used for object detection, then spatial coverage is improved, but feature extraction and object identification capability becomes inadequate
Solution Approach 1:
The patent segments the returned light packet into multiple volumetric components by sampling at different frequencies. This segmentation allows the system to extract detailed features from each component, transforming the 3D point cloud from a simple spatial map into a rich feature database. The volumetric segmentation enables simultaneous spatial coverage and detailed feature extraction, resolving the contradiction between broad coverage and measurement capability.
Solution Approach 2:
The patent adds a temporal dimension to the 3D point cloud by incorporating multi-frequency sampling data. This transforms the static spatial points into dynamic, feature-rich volumetric representations that contain both spatial and feature information. The additional temporal dimension enables the system to maintain comprehensive spatial coverage while extracting meaningful features from the returned light packets.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach increases device throughput, computes a higher number of points per emitted light packet, and provides additional feature elements, enabling more accurate and detailed object detection and feature extraction in various environmental conditions.
Implementation Method 1
LiDAR systems utilize one or more emitters and a detector array to sample incoming signal intensity at a pre-determined sampling frequency that generates two or more samples per emitted light packet
Implementation Method 2
Using a time of flight calculation applied to any reflections received, instead of just a phase shift analysis, the LiDAR unit can obtain range measurements
Implementation Method 3
The predecessor technology to current LiDAR units were object detection systems that could sense the presence or absence of objects within the field of view of one or more light beams based on phase shift analysis of the reflect light beam
Data Source
AI summary
LiDAR (light detection and ranging) systems use one or more emitters and a detector array to cover a given field of view where the emitters each emit a single pulse or a multi-pulse packet of light that is sampled by the detector array. On each emitter cycle the detector array will sample the incoming signal intensity at the pre-determined sampling frequency that generates two or more samples per emitted light packet to allow for volumetric analysis of the retroreflected signal portion of each emitted light packet as reflected by one or more objects in the field of view and then received by each detector.