Radar Reference Map Generation Using Gaussian Attributes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current radar localization systems face challenges in generating accurate and complete maps for autonomous vehicle operations, leading to increased driver takeovers and decreased safety and satisfaction due to insufficient or poor-quality radar reference maps.
Innovation Solution
The method involves receiving a radar occupancy grid, determining radar attributes based on occupancy probabilities, forming radar reference map cells, and generating a radar reference map by calculating Gaussians for cells with multiple attributes, which includes mean and covariance values, to create a robust and accurate spatial representation of the environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If radar reference maps are generated using traditional methods, then the generation process is simple, but the map quality is poor and incomplete leading to insufficient localization objects
Solution Approach 1:
The patent segments the map generation process into distinct stages: generating multiple candidate maps from radar data, evaluating each candidate map's quality metrics, and selecting the best map. This segmentation allows for comprehensive quality assessment without overwhelming complexity in a single step.
Solution Approach 2:
The patent performs preliminary actions by generating multiple candidate reference maps before final selection. Each candidate map is pre-evaluated using quality metrics to identify localization objects and assess completeness. This preliminary evaluation ensures high-quality map selection while maintaining a systematic generation process.
2Measurement precision
If more localization objects are included in the reference map, then localization accuracy improves, but the data processing complexity and time increase
Solution Approach 1:
The patent generates multiple candidate maps (excessive action) rather than a single map, allowing the system to select the best candidate that provides sufficient localization objects. This approach ensures high localization accuracy while avoiding the time cost of processing all possible maps, as only the best candidate is selected for use.
Solution Approach 2:
The patent replaces manual or simple automated map generation with an intelligent evaluation system that automatically assesses candidate maps using quality metrics. This substitution enables efficient processing of multiple candidates without linearly increasing time consumption, as the evaluation system quickly identifies the best map based on predefined criteria.
3Reliability
If the reference map is updated frequently to maintain accuracy, then localization reliability improves, but the computational load and processing time increase
Solution Approach 1:
The patent performs preliminary evaluation of candidate maps using quality metrics before final selection and deployment. This preliminary action ensures that only high-quality, reliable maps are selected for localization tasks, maintaining localization reliability while avoiding the computational waste of processing and updating low-quality maps frequently.
Solution Approach 2:
The patent implements a feedback mechanism where the quality evaluation metrics assess the performance and completeness of candidate maps. This feedback guides the selection process, ensuring that maps with sufficient localization objects are chosen. The feedback system allows the patent to maintain reliability by selecting maps that meet quality thresholds without requiring frequent updates, thus reducing computational load.
Data Source
AI summary
Methods and systems are described that enable radar reference map generation. A radar occupancy grid is received, and radar attributes are determined from occupancy probabilities within the radar occupancy grid. Radar reference map cells are formed, and the radar attributes are used to determine Gaussians for the radar reference map cells that contain a plurality of the radar attributes. A radar reference map is then generated that includes the Gaussians determined for the radar referenced map cells that contain the plurality of radar attributes. By doing so, the generated radar reference map is accurate while being spatially efficient.