Autonomous Vehicle Trajectory Standards via Interpretable Decision Trees
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in formalizing vehicle trajectory standards that are complex and difficult to interpret, as they often involve machine learning networks, making it hard to derive safe and comfortable driving behaviors analytically.
Innovation Solution
An inference system is implemented to infer vehicle trajectory standards using decision trees generated from training data, which are interpretable and explainable, allowing for the selection of vehicle trajectories that adhere to safety and comfort criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If machine learning networks are used to determine vehicle trajectory standards, then the ability to handle complex driving scenarios is improved, but interpretability and analytical derivability of safe driving behaviors deteriorate
Solution Approach 1:
The patent introduces decision trees as an intermediary between machine learning networks and trajectory standard formulation. The system uses labeled trajectory data from machine learning models to train decision tree models, which then generate interpretable trajectory standards. This intermediary layer preserves the adaptability of machine learning while achieving the interpretability needed for formal safety standards.
2Measurement precision
If complex machine learning models are used for trajectory prediction, then prediction accuracy is improved, but the complexity of deriving analytical safety criteria worsens
Solution Approach 1:
The patent creates simplified copies of complex machine learning behavior through decision tree models. Instead of directly analyzing complex neural network outputs, the system trains decision trees on labeled trajectory data to replicate the essential safety patterns. These decision tree copies are much easier to analyze analytically while preserving the predictive accuracy needed for safety criteria derivation.
3Manufacturing precision
If more detailed trajectory data is collected for training, then the precision of inferred standards is improved, but the amount of data processing and computation increases
Solution Approach 1:
The patent performs preliminary labeling of trajectory data with safety outcomes during the data collection phase. By pre-labeling trajectories as safe or unsafe based on expert analysis or simulation results, the system prepares the data in advance for efficient decision tree training. This preliminary action reduces the computational burden during the standard inference phase while maintaining high precision in the final standards.
Data Source
AI summary
Provided are methods, systems, and computer program products for inferring and using vehicle trajectory standards. An example method may include: obtaining a training dataset associated with an autonomous vehicle action, the training dataset comprising trajectory data for a plurality of examples of the autonomous vehicle action and labels for each of the plurality of examples; generating a decision tree based on the trajectory data and the labels of the training dataset; determining a vehicle trajectory standard based on the decision tree; and communicating the vehicle trajectory standard to at least one autonomous vehicle, wherein the at least one autonomous vehicle uses the vehicle trajectory standard to select a vehicle trajectory for the at least one autonomous vehicle.