Probabilistic Occupancy Grid Control for Autonomous Vehicle Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle control systems face challenges in balancing the influence of new and outdated sensor measurement data, leading to potential hazards due to uncertain decision-making based on outdated information, while requiring significant computational resources.
Innovation Solution
A method utilizing a probabilistic occupancy grid that associates each cell with multiple obstacle hypotheses and probability scores, updating these scores based on new measurement data and evolving them over time, with computational efficiency to manage resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the occupancy grid continuously updates probability scores for all cells using new sensor measurement data, then the reliability of vehicle control decisions is improved, but the computational load increases significantly
Solution Approach 1:
The occupancy grid is divided into two distinct regions: a first region (corresponding cells) where probability scores are updated using new sensor measurement data, and a second region (non-corresponding cells) where probability scores are evolved based on time elapsed since last update. This segmentation allows the system to focus computational resources on updating only the necessary portion of the grid while maintaining reliability across the entire occupancy grid.
2Device complexity
If the system uses outdated measurement data for vehicle control decisions, then the computational load is reduced, but the safety and reliability of vehicle control deteriorates
Solution Approach 1:
The system dynamically adjusts the treatment of probability scores based on the temporal relationship between sensor measurements and grid cells. For cells in the first region with corresponding sensor data, the system performs active updates. For cells in the second region without corresponding data, the system evolves probability scores over time based on elapsed time since last update. This dynamic approach ensures that control decisions are based on the most current relevant data while managing computational resources efficiently.
3Device complexity
If the occupancy grid evolves probability scores for all cells over time, then the system can operate with lower computational resources, but the measurement precision and accuracy of obstacle detection decreases
Solution Approach 1:
Different quality treatments are applied to different regions of the occupancy grid. In the first region (corresponding cells), the system applies high-quality updates using actual sensor measurement data to maintain high measurement precision for obstacle detection. In the second region (non-corresponding cells), the system applies lower-quality evolution based on time elapsed, which is sufficient for maintaining awareness of areas not currently observed by sensors. This local differentiation maintains overall detection accuracy while reducing computational load.
Data Source
AI summary
A method of controlling an autonomous vehicle which is movable on a surface, comprising: initializing an occupancy grid with reference map data, each cell in the occupancy grid being associated with a plurality of obstacle hypotheses and respective probability scores; repeatedly obtaining measurement data from sensors; associating the obtained measurement data with corresponding cells; in each of said corresponding cells, updating the probability score of at least one obstacle hypothesis in accordance with the obtained measurement data; outside said corresponding cells, evolving the probability scores based on the time elapsed since a latest update of these probability scores; and controlling the autonomous vehicle on the basis of the occupancy grid. In some embodiments, the probability score is updated based on a comparison of a measured elevation and an elevation indicated by the reference map data.