Partially-Learned Radar Tracking for Accurate Object Speed Estimates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Radar trackers in vehicle perception systems often fail to accurately and timely report object speeds, compromising situational awareness and safety in driving scenarios.
Innovation Solution
A partially-learned model that incorporates measured radial velocities into the radar processing pipeline, using a machine-learned model to analyze data cubes and fuse Doppler measurements with box predictions to improve speed estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a radar tracker uses complex data processing to improve speed estimation accuracy, then measurement precision improves, but processing time increases and reliability decreases
Solution Approach 1:
The patent segments the speed estimation process into two independent parts: (1) a machine-learned model that processes complex data cubes to provide box predictions and initial speed estimates, and (2) a traditional radar processor that independently extracts Doppler measurements from radar signals. This segmentation allows each component to optimize for its specific function without compromising the other, resolving the contradiction between accuracy and timeliness.
Solution Approach 2:
The patent merges the outputs of the machine-learned model and the traditional radar processor by fusing box predictions with Doppler measurements to produce final speed estimates. This combination allows the system to leverage both the pattern recognition capabilities of machine learning and the physical accuracy of Doppler-based measurements, achieving both high accuracy and reliability.
2Measurement precision
If a radar tracker uses machine learning models to improve object detection accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent divides the processing pipeline into distinct functional modules: a machine-learned model for pattern recognition and box prediction, and a traditional radar processor for physical measurement extraction. This segmentation reduces overall complexity by allowing each module to be optimized independently and simplifies debugging and maintenance.
Solution Approach 2:
The patent introduces an intermediary fusion mechanism that combines outputs from the machine-learned model and traditional radar processor. This intermediary layer acts as a mediator, integrating information from both sources without requiring direct integration of their complex processing logic, thereby reducing overall system complexity.
3Measurement precision
If a radar tracker processes all radar data to improve speed estimation accuracy, then measurement precision improves, but processing time increases
Solution Approach 1:
The patent segments the data processing workload by having the machine-learned model handle complex pattern recognition in parallel with the traditional radar processor handling Doppler extraction. This segmentation enables simultaneous processing of different data aspects, reducing total processing time while maintaining accuracy.
Solution Approach 2:
The patent applies partial action by having the machine-learned model provide box predictions and initial speed estimates that can be used immediately, while the traditional radar processor independently extracts Doppler measurements. This partial processing approach allows the system to deliver timely results even before complete data processing is finished, reducing overall processing time.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances the accuracy and reliability of speed estimates for objects, supporting safer and more comfortable vehicle operations without adding latency or hardware.
Implementation Method 1
determining, using a machine-learned model applied to the data cube, box predictions... Doppler measurements associated with groups of the potential detections associated with each of the box predictions are also determined... speed estimates for each of the box predictions by fusing the box predictions and the corresponding measured radial velocities
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This document describes techniques and systems for a partially-learned model for speed estimates in radar tracking. A radar system is described that determines radial-velocity maps of potential detections in an environment of a vehicle. The model uses a data cube to determine predicted boxes for the potential detections. Using the predicted boxes, the radar system determines Doppler measurements associated with the potential detections that correspond to the predicted boxes. The Doppler measurements are used to determine speed estimates for the predicted boxes based on the corresponding potential detections. These speed estimates may be more accurate than a speed estimate derived from the data cube and the model. Driving decisions supported by the speed estimates may result in safer and more comfortable vehicle behavior.