Autonomous Vehicle Reference Line Generation Using Multi-Thread Precalculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face challenges in efficiently generating smooth reference lines for motion planning, as existing optimization methods are time-consuming and can exceed the planning cycle time, limiting the length of reference lines that can be calculated and affecting navigation accuracy, especially on highways where longer planning horizons are required.
Innovation Solution
The system uses multiple processing threads to precalculate and store future reference lines based on predicted vehicle locations, allowing for concurrent generation and storage of multiple reference lines, which are then retrieved as needed, reducing the computational burden during planning cycles and ensuring sufficient time for trajectory generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If smooth optimization is performed on the reference line to improve navigation accuracy, then the quality of trajectory planning is improved, but the computation time increases and exceeds the planning cycle time
Solution Approach 1:
The system pre-calculates reference lines for future locations before the vehicle actually reaches those points. By performing the computationally intensive smooth optimization in advance (when computation resources are available), the system avoids time constraints during critical planning cycles. The pre-calculated reference lines are stored and retrieved when needed, ensuring both high accuracy and real-time performance.
2Length of moving object
If the reference line length is increased to extend planning horizon for highway driving, then the navigation accuracy and predictive capability are improved, but the optimization time increases proportionally
Solution Approach 1:
The system divides the long reference line into multiple segments corresponding to different future locations. Each segment is pre-calculated independently in advance, allowing the system to handle long planning horizons without proportionally increasing real-time computation time. When the vehicle reaches a specific location, only the relevant pre-calculated segment is retrieved, not the entire long reference line.
Solution Approach 2:
The reference line generation process is segmented by future location. Instead of calculating one long reference line at once, the system generates multiple shorter reference lines for different future points separately. This segmentation allows parallel processing and efficient memory management, reducing the time complexity from O(n) to O(1) for retrieval operations.
3Speed
If real-time reference line calculation is performed during planning cycle, then the system responsiveness is maintained, but the trajectory generation time is insufficient for complete optimization
Solution Approach 1:
The system performs reference line optimization in advance when computation resources are abundant, storing the results for rapid retrieval during real-time operation. This separates the computationally intensive optimization phase from the time-critical execution phase, allowing high-quality optimization without compromising real-time responsiveness.
Solution Approach 2:
Instead of recalculating the reference line during each planning cycle, the system creates copies of pre-calculated reference lines for future locations. These copies are stored in memory and quickly retrieved when needed, eliminating the need for repeated expensive optimization calculations while maintaining real-time performance.
Data Source
AI summary
Via a first processing thread, an ADV is controlled according to a first trajectory that was generated based on a first reference line starting at a first location. Concurrently via a second processing thread, a second reference line is generated based on a second location of the first trajectory that the ADV will likely reach within a predetermined period of time in future. The predetermined period of time is greater than or equals to an amount of time to generate a reference line for the ADV. The second reference line is generated while the ADV is moving according to the first trajectory and before reaching the second location. Subsequently, in response to determining that the ADV is within a predetermined proximity of the second location, a second trajectory is generated based on the second reference line without having to calculate the second reference line at the second location.