Autonomous Vehicle Trajectory Planning via Dynamic Conflict Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional autonomous vehicle control systems fail to accurately predict collisions with dynamic obstacles, particularly other vehicles, due to the lack of consideration for interaction between the vehicle and proximate dynamic vehicles, leading to suboptimal behavior and potential collisions.
Innovation Solution
A prediction-based system and method for trajectory planning that utilizes sensors to collect perception data, a computing device, and a trajectory prediction module to forecast the trajectories of other vehicles and dynamic objects, incorporating machine learning techniques to analyze real-world traffic data and predict their behavior, thereby modifying the vehicle's trajectory to avoid conflicts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional autonomous vehicle control systems use polynomial expressions or mathematical functions to represent spatial information of the driving path, then the vehicle can avoid stationary obstacles, but the system cannot accurately predict collisions with dynamic obstacles and fails to consider interactions with proximate dynamic vehicles
Solution Approach 1:
The system performs preliminary prediction of future positions and behaviors of proximate dynamic vehicles using machine learning models before making trajectory decisions. This advance prediction allows the autonomous vehicle to anticipate potential collisions and plan avoidance maneuvers proactively, rather than reactively responding to current obstacle positions.
Solution Approach 2:
The system continuously updates predictions of proximate vehicles' behaviors based on their actual movements and reactions to the autonomous vehicle's actions. This feedback loop refines the prediction model in real-time, improving accuracy of collision risk assessment and enabling adaptive trajectory planning that responds to dynamic changes in traffic interactions.
2Ease of operation
If the autonomous vehicle follows a computed driving path without considering proximate dynamic vehicles, then the navigation is simple, but the optimal behavior cannot be achieved and unexpected behaviors of proximate dynamic obstacles may result in collisions
Solution Approach 1:
The trajectory planning process is segmented into multiple stages: initial path computation using polynomial expressions, prediction of proximate vehicle behaviors using machine learning models, collision risk assessment based on predicted interactions, and trajectory optimization to avoid predicted collisions. This segmentation allows the system to maintain computational efficiency while incorporating complex interaction predictions.
3Measurement precision
If the system collects and processes perception data from sensors to predict trajectories of other vehicles, then the prediction accuracy improves, but the computational complexity and data processing requirements increase
Solution Approach 1:
The machine learning prediction model serves multiple functions: it predicts future positions of proximate dynamic vehicles, estimates their intended trajectories, assesses collision risks, and provides inputs for trajectory optimization. This multi-functionality reduces the need for separate specialized modules, thereby managing system complexity while maintaining high prediction precision across multiple task requirements.
Data Source
AI summary
A prediction-based system and method for trajectory planning of autonomous vehicles are disclosed. A particular embodiment is configured to: receive training data and ground truth data from a training data collection system, the training data including perception data and context data corresponding to human driving behaviors; perform a training phase for training a trajectory prediction module using the training data; receive perception data associated with a host vehicle; and perform an operational phase for extracting host vehicle feature data and proximate vehicle context data from the perception data, using the trained trajectory prediction module to generate predicted trajectories for each of one or more proximate vehicles near the host vehicle, generating a proposed trajectory for the host vehicle, determining if the proposed trajectory for the host vehicle will conflict with any of the predicted trajectories of the proximate vehicles, and modifying the proposed trajectory for the host vehicle until conflicts are eliminated.