Vehicle Route Crossing Prediction With Trajectory Error Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately predicting and controlling object crossings due to false positives and negatives in trajectory predictions, which can compromise safety and responsiveness.
Innovation Solution
The implementation of a machine learning model trained on predicted and ground truth trajectories, combined with timestamp information, to filter out false positives and negatives, allowing for enhanced prediction and control of object crossings by adjusting vehicle direction and speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models are used to predict object trajectories, then prediction capability is improved, but false positives and negatives increase reducing reliability
Solution Approach 1:
The system uses ground truth trajectory data to train and refine the machine learning model, creating a feedback loop where prediction errors are continuously reduced. The model learns from discrepancies between predicted and actual trajectories, improving both precision and reliability over time through iterative optimization.
Solution Approach 2:
The system performs preliminary filtering of trajectory predictions using multiple criteria including temporal consistency checks and spatial reasoning before making final crossing determinations. This preliminary action reduces false positives and negatives by eliminating obviously incorrect predictions before they affect the final decision.
2Reliability
If the vehicle adjusts direction and speed to control object crossings, then safety is improved, but vehicle maneuvering complexity increases
Solution Approach 1:
The system applies different control strategies to different objects based on their specific characteristics, trajectories, and potential crossing risks. Rather than uniform control adjustments, the vehicle tailors its maneuvering responses to local conditions, reducing overall control complexity while maintaining safety.
Solution Approach 2:
The system determines when partial control actions are sufficient to prevent crossings, rather than always applying maximum control. By using minimal necessary adjustments in direction and speed, the system maintains safety while reducing the complexity of control operations.
Data Source
AI summary
Provided are methods, systems and computer program products for predicting vehicle crossing and yielding, which can include receiving sensor information indicating an object surrounding a vehicle. Some methods also include determining a future position of the vehicle based on a first trajectory of the vehicle, determining a future position of the object based on a second trajectory of the object, and determining a vehicle control based on the future position of the vehicle and the future position of the object. The methods also include training a model using the vehicle control, the first trajectory of the vehicle, and the second trajectory of the object.