Radar Cross Traffic Tracking with Uncertainty-Aware Maneuver Risk Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle perception systems, particularly those using radar, face challenges with detection uncertainties and reduced information, leading to incomplete representations of uncertainties for maneuver risk assessments, which hinder accurate vehicle maneuver planning.
Innovation Solution
A risk maneuver assessment system that employs a processor-based controller with modules applying Markov Random Field algorithms, machine learning models, and adaptive thresholds to filter and track target objects, predict maneuver risks, and generate control commands, utilizing a combination of radar, lidar, and image sensor data for enhanced uncertainty representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If radar scan detections are used for target detection, then detection coverage is provided, but detection uncertainty and location uncertainty increase
Solution Approach 1:
The patent combines multiple radar detections across different scan cycles and integrates them with map data and tracking algorithms to form a unified target representation. This merging process reduces uncertainty by aggregating multiple noisy measurements into a single, more reliable target state estimate with associated covariance information.
Solution Approach 2:
The system implements feedback through iterative tracking where previous target state estimates and covariance information are fed back into the detection and tracking process. The uncertainty from previous scans is carried forward and updated with new measurements, allowing the system to progressively reduce uncertainty through repeated observations and feedback loops.
2Reliability
If multiple candidate detections are processed to identify true target, then detection completeness improves, but computational complexity increases
Solution Approach 1:
The patent transforms the target identification problem by changing parameters from raw detection coordinates to uncertainty-aware state representations including covariance information. By working in this transformed parameter space and using probabilistic metrics, the system can efficiently evaluate multiple candidates without exhaustive search, reducing computational complexity while maintaining accuracy.
Solution Approach 2:
The system replaces traditional mechanical or deterministic tracking methods with probabilistic and statistical approaches. Instead of using fixed threshold or rule-based filtering, the patent employs probability density functions and covariance-based uncertainty measures to automatically evaluate and select true targets from multiple candidates, reducing the need for complex manual processing rules.
3Productivity
If uncertainty representation is simplified for faster processing, then processing speed improves, but maneuver risk assessment accuracy deteriorates
Solution Approach 1:
The patent extracts and separates the essential uncertainty information (covariance matrices and probability densities) from the full detection data, carrying forward only the critical uncertainty parameters needed for risk assessment. This extraction allows the system to maintain accurate uncertainty representation while reducing the amount of data that must be processed in real-time, improving processing speed without sacrificing assessment accuracy.
Data Source
AI summary
A risk maneuver assessment system and method to generate a perception of an environment of a vehicle and a behavior decision making model for the vehicle; a sensor system configured to provide the sensor input in the environment for filtering target objects; one or more modules configured to map and track target objects to make a candidate detection from multiple candidate detections of a true candidate detection as the tracked target object; apply a Markov Random Field (MRF) algorithm for recognizing a current situation of the vehicle and predict a risk of executing a planned vehicle maneuver at the true detection of the dynamically tracked target; apply mapping functions to sensed data of the environment for configuring a machine learning model of decision making behavior of the vehicle; and apply adaptive threshold to cells of an occupancy grid for representing an area of tracking of objects within the vehicle environment.