Radar Observation Model Using Machine Learning for Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in collecting and interpreting environmental data for precise navigation, particularly in processing radar measurements from multiple points, which affects the accuracy of tracking and detecting dynamic objects.
Innovation Solution
A machine learning model, such as a neural network, is trained to receive radar measurement data and a predicted state associated with objects, generating improved track data by utilizing prior knowledge, similar to a Kalman filtering process, to simplify processing and enhance precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional radar measurement processing methods are used, then processing complexity is high, but measurement precision and tracking accuracy are limited
Solution Approach 1:
The patent replaces traditional mechanical/mathematical radar processing systems with a neural network-based machine learning system. The neural network is trained to directly map radar measurement inputs to tracking outputs, substituting complex iterative algorithms with a trained model that provides both high accuracy and efficient processing.
Solution Approach 2:
The patent transforms the processing approach by changing from deterministic algorithmic parameters to learned parameters through neural network training. The system learns optimal processing parameters from training data, enabling adaptive parameter adjustment that improves tracking accuracy while reducing computational complexity during operation.
2Measurement precision
If multiple radar measurement points are processed in detail, then tracking precision improves, but processing time and computational load increase
Solution Approach 1:
The patent performs preliminary action by training the neural network offline before deployment. During real-time operation, the pre-trained network quickly processes radar measurements without requiring complex computations, thus reducing processing time while maintaining high tracking precision through the learned patterns from training data.
3Reliability
If complex radar observation models are used, then tracking accuracy improves, but system complexity and computational requirements increase
Solution Approach 1:
The patent substitutes complex analytical radar observation models with a neural network-based model. The neural network learns the complex relationships between radar measurements and target states from training data, achieving high tracking reliability without requiring explicit mathematical models, thereby reducing system complexity and computational requirements during operation.
Data Source
AI summary
A method includes obtaining a first track associated with a first time. A first track associated with a first time is obtained. First predicted state data associated with a second time that is later than the first time, are generated based on the first track. Radar measurement data associated with the second time are obtained from one or more radar sensors. Track data are generated by a machine learning model based on the first predicted state data and the radar measurement data. Second predicted state data associated with the second time are generated based on the first track. A second track associated with the second time is generated based on the track data and the second predicted state data. The second track associated with the second time is provided to an autonomous vehicle control system for autonomous control of a vehicle.