Autonomous Vehicle Three-Point-Turn Planning via Cost Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Planning and executing a three-point-turn for autonomous driving vehicles is complex due to the need for precise navigation and cost assessment, as it differs significantly from normal routing operations, and existing methods do not effectively account for obstacles and remaining lane length in determining feasibility.
Innovation Solution
A computer-implemented method for planning a three-point-turn in autonomous driving vehicles, which involves determining a candidate route, categorizing it into segments, calculating a total cost incorporating obstacle and remaining lane length costs, and planning the turn based on whether the cost is below a threshold, using algorithms like A-Star search to assess feasibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a three-point-turn is planned using normal routing methods, then the routing process is simple, but the maneuver cannot be safely executed due to lack of consideration for obstacles and lane length constraints
Solution Approach 1:
The patent divides the routing process into two distinct modes: normal routing for standard navigation and three-point-turn routing for reversal maneuvers. This segmentation allows each mode to be optimized independently, with three-point-turn routing incorporating specialized cost assessments for obstacles and lane length without complicating normal routing operations.
Solution Approach 2:
The patent implements dynamic switching between normal routing and three-point-turn routing modes based on the maneuver type detected. The system adapts its routing approach in real-time, applying the appropriate cost function and planning algorithm depending on whether the vehicle needs to perform a standard navigation task or a three-point-turn reversal.
2Reliability
If the total cost threshold is set low to ensure safety, then obstacle avoidance is improved, but the vehicle may fail to execute valid three-point-turns due to overly restrictive criteria
Solution Approach 1:
The patent changes the cost function parameters dynamically based on the routing mode. For three-point-turn routing, the cost function incorporates obstacle costs and remaining lane length costs with specific weights that reflect the maneuver's requirements. This parameter adjustment allows the system to evaluate three-point-turn feasibility accurately without using overly conservative thresholds that would prevent valid maneuvers.
Solution Approach 2:
The patent applies different cost assessment criteria to different segments of the routing problem. Normal routing segments use standard cost functions, while three-point-turn segments use specialized cost functions that account for obstacles and lane length. This localized quality approach ensures that safety considerations are applied where needed without unnecessarily restricting overall maneuver execution.
3Measurement precision
If the cost function includes both obstacle cost and remaining lane length cost, then the accuracy of feasibility assessment is improved, but the computational complexity increases
Solution Approach 1:
The patent segments the cost calculation into distinct components: obstacle cost, remaining lane length cost, and total cost. Each component is calculated separately using dedicated functions, which improves measurement precision by ensuring all relevant factors are considered. The segmented structure also makes the complexity manageable by organizing calculations into modular, independent modules.
Solution Approach 2:
The patent creates a universal cost function framework that can handle both normal routing and three-point-turn routing through a unified interface. The cost function module serves multiple purposes: it calculates costs for standard navigation, evaluates three-point-turn feasibility, and provides the basis for mode switching decisions. This multi-functionality reduces overall system complexity despite the enhanced assessment capabilities.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
A three-point-turn is planned and executed in the operation of an autonomous driving vehicle (ADV). A candidate route from a start point and going through an end point is determined, the start point and the end point being in lanes associated with opposite travel directions. The candidate route is categorized into partially overlapping first, second, and third segments. A total cost associated with the candidate route is determined based at least in part on the first and second segments. Whether the total cost is below a threshold cost is determined. In response to a determination that the total cost is below the threshold cost, the three-point-turn is planned based on the candidate route. Further, driving signals are generated based at least in part on the planned three-point-turn to control operations of the ADV.