Hybrid Trajectory Prediction for Vehicle Interaction Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for determining the expected trajectory of objects, such as vehicles or pedestrians, in partially autonomous vehicles lack efficiency and reliability, particularly in modeling interactions with other vehicles and accurately predicting long-term movements.
Innovation Solution
A computer-implemented method that combines machine-learning and model-based methods to predict or reconstruct trajectories, using sensor data like radar, lidar, and camera data, by fusing intermediate data from both approaches through a neural network-based fusion network, which incorporates learnable parameters to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a model-based method is used to predict trajectory, then the prediction follows physical laws, but it cannot accurately describe complex interactions with other vehicles
Solution Approach 1:
The patent combines model-based prediction (providing physical law compliance) with machine-learning-based prediction (providing interaction modeling capability) into a unified hybrid system. The model-based component ensures predictions follow physical laws, while the machine-learning component learns complex interaction patterns from data, together resolving the contradiction between reliability and adaptability.
Solution Approach 2:
The prediction system uses a composite approach by integrating two different prediction methodologies (model-based and machine-learning-based) into a single hybrid system. This composite structure allows the system to leverage the strengths of both approaches: the physical correctness of model-based methods and the adaptive interaction modeling of machine learning.
2Adaptability or versatility
If machine-learning method is used to learn complex interactions, then the interaction modeling improves, but the system cannot fully replace known physical models
Solution Approach 1:
The patent merges machine-learning-based prediction with model-based prediction in a hybrid system. The machine-learning component handles complex interaction modeling while the model-based component ensures physical law compliance, together resolving the contradiction between adaptability and reliability.
Solution Approach 2:
The hybrid system acts as an intermediary that combines the outputs or intermediate representations from both machine-learning and model-based methods. This intermediary structure allows the system to leverage both approaches without fully replacing either, maintaining reliability while improving interaction modeling.
3Device complexity
If only model-based method is used, then the system is simpler, but the long-term trajectory prediction accuracy deteriorates
Solution Approach 1:
The patent combines model-based and machine-learning-based prediction methods into a hybrid system that achieves superior long-term trajectory prediction accuracy. The integration allows the system to capture both physical laws and complex interaction patterns, resolving the contradiction between simplicity and accuracy.
4Adaptability or versatility
If only machine-learning method is used, then the interaction modeling improves, but the system loses interpretability and physical law compliance
Solution Approach 1:
The patent merges machine-learning-based prediction with model-based prediction, where the model-based component provides interpretability and physical law compliance while the machine-learning component provides interaction modeling capability. This combination resolves the contradiction between adaptability and interpretability.
Data Source
AI summary
A method for determining information on an expected trajectory of an object comprises: determining input data being related to the expected trajectory of the object; determining first intermediate data based on the input data using a machine-learning method; determining second intermediate data based on the input data using a model-based method; and determining the information on the expected trajectory of the object based on the first intermediate data and based on the second intermediate data.