Imaging Radar Fusion for Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multiple-object tracking systems face challenges such as frequent occlusions, initialization and termination of tracks, changes in object appearances, and interactions among objects, which complicate reliable identification and prediction of object paths, especially in adverse weather conditions.
Innovation Solution
The integration of imaging sensor and radar fusion using tracking-by-detection methods, where a supervised learning neural network calculates probabilities of object continuity across frames from both sensors, and these probabilities are fused to enhance tracking accuracy and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If only imaging sensor data is used for tracking, then the system is simpler, but tracking reliability deteriorates under occlusions and adverse weather
Solution Approach 1:
The patent combines imaging sensor data and radar sensor data into a unified tracking system. The neural network processes inputs from both sensor types to generate track hypotheses, leveraging the complementary strengths of each sensor to maintain reliable tracking under occlusions and adverse weather conditions where single-sensor systems fail.
2Measurement precision
If multiple sensors are fused for tracking, then tracking accuracy improves, but computational complexity increases
Solution Approach 1:
The tracking system is segmented into distinct functional modules: an imaging sensor module, a radar sensor module, a neural network processing module that generates track hypotheses, and a probability calculation module. This segmentation allows each component to specialize in specific tasks, improving overall tracking accuracy while managing computational complexity through modular design.
Solution Approach 2:
The system calculates probabilities for multiple track hypotheses beyond what a single-sensor system would generate. By evaluating multiple potential tracks and their associated probabilities, the system achieves higher tracking accuracy through selective refinement of the most promising hypotheses rather than exhaustively processing all possibilities.
3Reliability
If tracking-by-detection is used with neural networks, then object identification improves, but processing time increases
Solution Approach 1:
The neural network performs preliminary processing of sensor data to generate track hypotheses and probability estimates before final track confirmation. By pre-calculating likelihoods for potential object identities and trajectories, the system reduces the computational burden of final decision-making and accelerates the overall processing pipeline.
Data Source
AI summary
This document describes methods and systems directed at imaging sensor and radar fusion for multiple-object tracking. Using tracking-by-detection, an object is first detected in a frame captured by an imaging sensor, and then the object is tracked over several consecutive frames by both the imaging sensor and a radar system. The object is tracked by assigning a probability that the object identified in one frame is a same object identified in the consecutive frame. A probability is calculated for each data set captured by a sensor by a supervised-learning neural-network model using the data collected from the sensors. Then, the probabilities associated with each sensor are fused into a refined probability. By fusing the data gathered by the imaging sensor and the radar system in the consecutive frames, a safety system can track multiple objects more accurately and reliably than using the sensor data separately to track objects.