Multi-Sensor Occupancy Mapping for Occluded Surrounding Areas
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems face challenges in generating accurate occupancy maps, particularly in areas occluded by objects, where sensors have overlapping sensing ranges but capture different data due to their distinct positions, leading to incomplete point cloud data.
Innovation Solution
A processor-implemented method that generates an occupancy map by supplementing occluded area information in the first point cloud data using corresponding portions of the second point cloud data, based on temporal adjacency and localization information, determining voxel values with weights based on sensor distances, and selecting the appropriate second sensor for data capture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single sensor is used for occupancy map generation, then the device complexity is reduced, but the measurement precision deteriorates due to occluded areas
Solution Approach 1:
The patent combines point cloud data from multiple sensors (first sensor and second sensor) to generate a comprehensive occupancy map. The processor integrates data from both sensors, using the second sensor's data to supplement occluded areas detected by the first sensor, thereby improving occupancy map accuracy while managing system complexity through coordinated multi-sensor operation.
2Measurement precision
If multiple sensors with overlapping sensing ranges are deployed, then the measurement precision improves by covering occluded areas, but the device complexity increases
Solution Approach 1:
The patent segments the sensing task by assigning different sensors to different spatial zones. The first sensor covers the primary sensing range, while the second sensor covers the overlapping area to capture occluded regions. This segmentation allows each sensor to specialize in specific areas, improving overall measurement precision without requiring a single complex sensor system.
Solution Approach 2:
The processor acts as an intermediary that receives data from both sensors, processes the point cloud information, and generates the final occupancy map. It mediates between the multiple sensors by integrating their data, resolving the complexity of coordinating multiple sensors through a centralized processing unit that harmonizes their outputs.
3Measurement precision
If point cloud data from temporally adjacent timepoints is used, then the occupancy map accuracy improves by compensating for temporal changes, but the processing complexity increases
Solution Approach 1:
The system performs preliminary actions by collecting point cloud data from multiple timepoints before generating the final occupancy map. It prepares temporal data from the first and second sensors at different acquisition timepoints, then processes this pre-collected temporal information to compensate for changes occurring between timepoints, improving temporal accuracy without requiring complex real-time processing during map generation.
Data Source
AI summary
A processor-implemented method includes, based on an overlapping area between a first sensing range of a first sensor and a second sensing range, of a second sensor, that includes an area occluded to the first sensor by an object in the first sensing range, generating an occupancy map by supplementing information about the occluded area among the first point cloud data based on a corresponding portion of the second point cloud data, wherein an acquisition timepoint of the first point cloud data and an acquisition timepoint of the second point cloud data are a same timepoint or temporally adjacent to each other, wherein the first point cloud data corresponds to the first sensing range of the first sensor from a perspective of the first sensor of a moving object, where the first sensing range corresponds to a surrounding area of the moving object, and wherein the second point cloud data corresponds to the second sensing range of the second sensor that is remote from the first sensor, where the second sensing range corresponds to a surrounding area of the second sensor.