Polar Occlusion Maps for Autonomous Vehicle Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting and tracking objects by autonomous vehicles lack sufficient speed, precision, and accuracy, leading to inadequate detection of objects proximate to the vehicle.
Innovation Solution
A computer-implemented method that receives sensor data, determines inflection points where objects occlude detection, and generates sparse map data using polar coordinates to create occlusion maps, which are then used to improve object detection and tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing object detection methods are used, then the system can detect objects in the environment, but the detection lacks sufficient speed, precision, and accuracy
Solution Approach 1:
The patent segments the environment into occluded and non-occluded regions using polar occlusion maps. By dividing the detection space and processing only relevant segments (occlusion boundaries rather than entire point clouds), the system achieves both high precision in detecting occluded objects and improved processing speed through selective analysis.
Solution Approach 2:
The patent transforms the detection problem by introducing polar coordinates (radial distance and angular position) as an additional dimensional framework. This dimensional change enables more precise representation of occlusion boundaries and improves detection accuracy while maintaining computational efficiency through geometric transformations.
2Measurement precision
If detailed sensor data is processed to improve detection accuracy, then precision increases, but computational resources and bandwidth usage increase
Solution Approach 1:
The patent extracts only the essential occlusion boundary information from complete sensor data by identifying inflection points in polar coordinates. This extraction approach achieves high detection accuracy by focusing on critical occlusion features while significantly reducing computational resource consumption and bandwidth usage compared to processing full point cloud data.
Solution Approach 2:
The patent changes the data representation parameters from Cartesian coordinates to polar coordinates, and from complete point cloud data to sparse inflection point sets. This parameter transformation maintains detection accuracy by preserving occlusion boundary information while reducing data volume and computational requirements for processing.
Data Source
AI summary
Systems, methods, tangible non-transitory computer-readable media, and devices for operating an autonomous vehicle are provided. For example, a vehicle computing system can receive sensor data including information based on sensor outputs associated with detection of objects in an environment by sensors of a vehicle. Inflection points can be determined based at least in part on the sensor data. The inflection points can correspond to portions of the objects that occlude detection of the environment beyond the portions of the objects. A set of polar coordinates can be determined for each of the one or more inflection points. The set of polar coordinates can include a distance from the sensors to a portion of the objects and an angle of the sensor with respect to the portion of the objects. Furthermore, sparse map data including the set of polar coordinates can be generated based on the set of polar coordinates.