Intersection Classification via Fleet Sensor Trajectories
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated vehicle systems face challenges in navigating urban traffic intersections due to reliance on incomplete or incorrect classification of right-of-way rules and intersection types, which are often derived from faulty or missing traffic control elements in road signage and infrastructure.
Innovation Solution
A computing system that classifies road network intersections using sensor data from human-driven vehicles, processing trajectories and behaviors to generate vehicle traces and classify intersections, thereby deriving intersection types and rules independently of traffic signs and infrastructure, using both heuristic and learning-based approaches to improve autonomous vehicle navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If right-of-way rules and intersection types are derived from road signage or traffic control elements in recorded ground truth maps, then the classification process is straightforward, but faulty or missing labels lead to incorrect classification
Solution Approach 1:
The system uses sensor data from fleet vehicles to automatically classify intersections and update ground truth maps without requiring manual human labeling. The vehicles themselves generate the data needed to improve the system, making the map creation process self-sustaining and eliminating human error in labeling while maintaining high reliability through objective sensor-based classification
Solution Approach 2:
The system continuously collects sensor data from fleet vehicles, processes it to classify intersections, and uses this information to update and improve ground truth maps. This creates a feedback loop where real-world vehicle behavior data continuously refines the classification accuracy, allowing the system to correct errors in existing maps while maintaining ease of updates
2Device complexity
If automated vehicle systems use limited parameters for decision-making, then the system complexity is reduced, but the system cannot handle complex urban traffic intersection scenarios
Solution Approach 1:
The system segments the complex task of intersection navigation into distinct components: sensor data collection from fleet vehicles, trajectory processing, intersection classification, and rule derivation. Each component handles a specific aspect of the problem, maintaining manageable system complexity while collectively enabling comprehensive urban intersection navigation capability
Solution Approach 2:
The system adds a new dimension to automated vehicle decision-making by incorporating classified intersection types and derived right-of-way rules from ground truth maps. This transforms the decision-making process from simple parameter-based reactions to context-aware navigation that understands intersection geometry and traffic control requirements, significantly enhancing adaptability to complex urban scenarios
Data Source
AI summary
A system can perform a method that includes receiving sensor data from a subset of human-driven vehicles that have moved through an intersection in a region, where the sensor data indicates a respective set of trajectories of respective human-driven vehicles of the subset through the intersection. The method can further include processing the sensor data to classify driving behavior of each human-driven vehicle of the subset of human-driven vehicles through the intersection. The method can further include classifying the intersection to label an autonomy map to include pass-through information for at least one of autonomous vehicles or semi-autonomous vehicles driving through the intersection.