Trajectory Collision Risk Assessment via Grid Maps and Driving Envelopes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory planning schemes in autonomous driving face inaccuracies in obstacle representation and collision risk estimation due to reliance on polygon-based obstacle segmentation and coarse vehicle center point estimation, leading to suboptimal trajectory decisions.
Innovation Solution
The method processes environment grid maps to generate occupancy and cost grid maps, representing obstacle information densely and calculating collision risk based on a vehicle's driving envelope along candidate trajectories, avoiding polygon segmentation and enhancing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If polygon-based obstacle segmentation is used, then obstacle representation is simplified, but accuracy loss occurs in obstacle information
Solution Approach 1:
The patent uses grid maps as a simplified copy representation of the environment, where each grid cell stores occupancy probability and cost information. This copying approach avoids the need for complex polygon segmentation while preserving essential obstacle information through dense grid-based representation, resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The patent segments the environment into discrete grid cells rather than using continuous polygon representations. This grid segmentation allows for dense obstacle information storage in each cell while maintaining computational simplicity, eliminating the accuracy loss associated with polygon-based segmentation.
2Device complexity
If vehicle center point estimation is used, then collision risk calculation is simplified, but estimation accuracy deteriorates
Solution Approach 1:
The patent transitions from one-dimensional vehicle center point estimation to two-dimensional driving envelope representation. The driving envelope captures the full spatial extent of the vehicle's possible positions, providing comprehensive collision risk assessment across the entire vehicle footprint rather than relying on a single center point, thus improving accuracy without excessive complexity.
Solution Approach 2:
The patent changes the parameter representation from a single point (vehicle center) to a distributed set of points forming an envelope. By representing the vehicle's spatial occupation as a collection of grid cells within the driving envelope, the system achieves more accurate collision risk estimation while maintaining computational efficiency through grid-based operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to the technical field of autonomous driving. Provided are a trajectory collision risk assessment method and apparatus, and an electronic device and a storage medium. The trajectory collision risk assessment method comprises: processing an environment grid map on the basis of a driving region of interest, so as to obtain an occupancy grid map, which at least represents obstacle information of the driving region of interest; processing the occupancy grid map according to the obstacle information, so as to obtain a cost grid map, which at least represents a collision risk of the driving region of interest; obtaining a candidate trajectory set, and generating a driving envelope of a target vehicle moving along each candidate trajectory; and determining a collision risk cost of each candidate trajectory according to the driving envelope and the cost grid map. In the trajectory collision risk assessment solution of the present invention, obstacle information is densely expressed on the basis of a grid map, such that a precision loss caused by point cloud segmentation and clustering can be avoided, and on the basis of a driving envelope of a vehicle contour, a collision risk cost of a candidate trajectory can be accurately calculated.