Inferred Stop Line Positioning for Unpainted AV Intersections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in determining the correct stopping location at intersections without painted stop lines, leading to unnatural stops, potential traffic conflicts, and obstructed line-of-sight issues.
Innovation Solution
A system that uses sensors and high-definition map data to infer a stop line by identifying a restricted lane segment and advancing an entry line towards a nearest lane conflict or crosswalk, forming an inferred stop line at a target distance, allowing the vehicle to decelerate and stop accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the vehicle stops at the painted stop line, then the stopping location is natural and traffic flow is smooth, but the stop line may be absent or undetectable
Solution Approach 1:
The system pre-processes map data to identify intersections and determine expected stop line locations before the vehicle reaches them. By preparing reference stop line positions in advance based on map features (intersections, traffic control measures), the system has fallback positions ready when painted stop lines are undetectable, ensuring reliable stopping without waiting for potentially missing visual cues.
Solution Approach 2:
The system introduces map data as an intermediary between the vehicle's vision system and the stopping decision. When the vision system fails to detect painted stop lines, the map data provides inferred stop line positions based on intersection geometry and traffic control measure locations, acting as a mediator to bridge the detection gap and guide the vehicle to appropriate stopping positions.
2Reliability
If the vehicle stops farther back from the intersection, then the vehicle has better visibility, but other drivers may proceed out of turn and traffic conflicts increase
Solution Approach 1:
The system dynamically adjusts the stopping position parameter based on multiple factors including intersection type, presence of crosswalks, traffic control measures, and detected obstructions. By changing the stop line position parameter adaptively rather than using a fixed distance rule, the system optimizes the balance between visibility requirements and traffic flow maintenance, preventing both premature stops and excessive backing up.
Solution Approach 2:
The system makes the stopping position dynamic rather than static. It continuously evaluates map data about intersection geometry, traffic control measures, and environmental obstructions to determine the optimal stop line position for each specific context. This dynamic adjustment allows the vehicle to stop at the furthest position that still maintains proper traffic order, adapting to varying visibility conditions without causing traffic conflicts.
3Productivity
If the vehicle stops at the correct location, then traffic flow is maintained, but the vehicle may encounter obstructions blocking line-of-sight
Solution Approach 1:
The system pre-identifies potential obstruction locations using map data about buildings and other structures near intersections. By knowing where obstructions are located in advance, the system can proactively adjust the inferred stop line position to ensure the vehicle stops at a location with adequate visibility, preventing line-of-sight blockage before the vehicle even approaches the intersection.
4Measurement precision
If the system uses complex inference algorithms, then stopping accuracy improves, but computational complexity increases
Solution Approach 1:
The inference system is divided into separate modular components: map data processing module, intersection detection module, stop line inference module, and obstruction handling module. Each module performs a specific function independently, processing map features and combining results to determine the final stop position. This segmentation allows the complex inference task to be broken into manageable pieces that can be executed efficiently and maintained independently.
Data Source
AI summary
A system and method for inferring a stop line for a vehicle at an intersection are provided. The system includes a processor configured to detect from accessed map data that a traffic control measure is positioned before an intersection and determine whether a stop line for the detected traffic control measure is painted. The processor, in response to determining that no stop line is painted, identifies a restricted lane and infers a stop line. The processor infers the stop line by identifying, as a nearest lane conflict, a lane segment of a second road intersecting the first road at the intersection and advancing a location of the entry line as an intermediate stop line a distance toward the nearest lane conflict, until the intermediate stop line is at a target distance from a nearest boundary of the nearest lane conflict to form an inferred stop line.