Trajectory Assessment via Recommendation Service for Automated Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current trajectory planning systems for automated vehicles face limitations in handling the vast number of possible traffic scenarios, leading to overly cautious or dangerous decisions due to the combinatorial explosion of alternatives and the difficulty in processing numerous calculated trajectories, which are not adequately covered by existing classification models and machine learning approaches.
Innovation Solution
A method utilizing a recommendation service to assess possible trajectories of road users by capturing static and dynamic features of the traffic environment, identifying road users, and determining their trajectories using collaborative filtering and neural networks, allowing for accurate and efficient prediction of user behavior without the need for explicit rule development or extensive training data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a fixed number of trajectory classes are used for classification, then the system complexity is reduced and processing becomes easier, but not all traffic situation options are covered leading to overly cautious or dangerous planning
Solution Approach 1:
The patent transitions from static, pre-defined trajectory classes to dynamic, continuous trajectory generation. The system now adapts trajectory possibilities based on real-time traffic situations, road user behaviors, and environmental conditions, allowing the trajectory space to expand or contract dynamically rather than being fixed in advance.
Solution Approach 2:
The system changes the parameterization approach from discrete class labels to continuous trajectory parameters. Instead of selecting from a fixed set of classes, the system generates trajectories by varying parameters such as position, velocity, acceleration, and timing, enabling fine-grained differentiation of trajectory options without increasing system structural complexity.
2Reliability
If the number of trajectory classes is increased to cover all possible scenarios, then trajectory planning reliability improves, but the system reaches functional limits due to the large number of possibilities and combinatorial explosion
Solution Approach 1:
The patent segments the trajectory generation process into modular components: road user detection, behavior modeling, trajectory prediction, and assessment. Each component handles a specific aspect independently, avoiding the need to manage all trajectory possibilities in a single complex classification system. This modular segmentation enables comprehensive coverage without overwhelming system complexity.
Solution Approach 2:
The system introduces intermediate representations such as behavior models and probability distributions that mediate between raw sensor data and final trajectory decisions. These intermediaries process and structure information in manageable forms, preventing the combinatorial explosion that would occur with direct classification of all possible trajectories.
3Adaptability or versatility
If machine learning approaches are used to estimate agent trajectories, then the system can handle complex patterns, but the large number of calculated trajectories is difficult to process further without assessing individual trajectories
Solution Approach 1:
The system performs preliminary assessment of trajectory possibilities by evaluating probability distributions and filtering out highly improbable trajectories before detailed processing. This preliminary filtering reduces the number of trajectories requiring full assessment, improving processing efficiency while maintaining adaptability through the probability-based approach.
Solution Approach 2:
The system applies partial assessment to trajectory sets, focusing computational resources on the most probable and relevant trajectories rather than assessing all possible trajectories in equal detail. This selective assessment approach maintains high adaptability for critical trajectories while improving overall processing efficiency by reducing unnecessary computations.
Data Source
AI summary
A method for assessing possible trajectories of road users in a traffic environment includes capturing the traffic environment with static and dynamic features, identifying at least one traffic user, determining at least one possible trajectory for at least one road user in the traffic environment, and assessing the at least one determined possible trajectory for the at least one road user with an adapted/trained recommendation service and the captured traffic environment.