Curvature Corrected Path Sampling for Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous driving systems face challenges in generating stable trajectories for autonomous vehicles, particularly in navigating curved paths, as they rely on reference lines that may not account for maximum curvature and turning capabilities, leading to inefficient path planning and potential safety issues.
Innovation Solution
A curvature-corrected path sampling system that determines a first trajectory based on map and route information, calculates a maximum curvature, and generates sample points with a sampling decay ratio to create a second trajectory that adjusts to the left or right of the first trajectory, considering the vehicle's speed and turning capabilities, thereby optimizing path generation and control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the system uses reference lines for trajectory planning, then the path following capability is improved, but the trajectory stability deteriorates due to inadequate consideration of maximum curvature and turning capabilities
Solution Approach 1:
The system dynamically adjusts the sampling decay ratio based on the maximum curvature of the reference line. When curvature increases, the sampling decay ratio changes to generate sample points that account for the vehicle's turning capabilities, ensuring trajectory stability while maintaining path following accuracy
Solution Approach 2:
The trajectory generation process is made dynamic by continuously evaluating the maximum curvature along the reference line and adjusting the sampling strategy accordingly. This allows the system to adapt to varying road conditions and maintain stable trajectories through curved sections
2Device complexity
If the system generates trajectories without curvature correction, then the computational complexity is reduced, but the path planning efficiency deteriorates due to inadequate consideration of vehicle turning capabilities
Solution Approach 1:
The system performs preliminary analysis by determining the maximum curvature of the reference line before generating sample points. This preliminary action allows the system to optimize the sampling decay ratio in advance, improving path planning efficiency without significantly increasing overall computational complexity
Solution Approach 2:
The sampling decay ratio is adjusted locally based on the maximum curvature characteristics of different road sections. This localized adaptation allows the system to improve path planning efficiency in curved sections while maintaining reasonable computational complexity in straight sections
3Ease of manufacture
If the system uses uniform sampling along the reference line, then the implementation simplicity is improved, but the trajectory accuracy deteriorates due to inadequate adaptation to curvature variations
Solution Approach 1:
The system changes the sampling parameter (decay ratio) based on the maximum curvature of the reference line. This parameter adaptation allows the system to achieve high trajectory accuracy in curved sections while maintaining the simplicity of the sampling framework
Solution Approach 2:
The sampling strategy transitions from static uniform sampling to dynamic adaptive sampling, where the sampling decay ratio adjusts according to the local curvature characteristics of the reference line, improving trajectory accuracy without significantly complicating the implementation
Data Source
AI summary
According to one embodiment, a system determines a first trajectory (e.g., a reference line) based on map and route information of the ADV. The system determines a maximum curvature for a predetermined distance for the first trajectory. The system determines a sampling decay ratio based on the maximum curvature. The system generates one or more sets of sample points for the first trajectory, where each set of sample points include one or more sample points separated by a lateral distance apart based on the sampling decay ratio. The system generates a second trajectory for the ADV based on the sets of sample points to control the ADV, where the second trajectory nudges to a left or a right of the first trajectory based on the maximum curvature of the first trajectory.