Radar Vision Fusion for Vehicle Target Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle detection systems in autonomous vehicles face challenges such as inaccuracy in longitudinal and lateral distance detection due to low visibility conditions, high rates of false detection, and failure in cluttered environments, which hinder the development of efficient Advanced Driver Assistance Systems (ADAS) and Autonomous Drive (AD) systems.
Innovation Solution
A system that combines radar and vision-based detection units in a host vehicle, using radar sensors and image sensors to identify and track targets by matching detected targets across both systems, categorizing them as 'locked targets' for consistent identification, and providing control signals for vehicle control, even when the target is not detected by the vision system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vision-based detection units are used to detect targets, then lateral distance detection accuracy is improved, but detection performance degrades in low visibility conditions
Solution Approach 1:
The patent combines radar-based detection units and vision-based detection units into a unified detection system. The radar unit provides reliable detection in low visibility conditions while the vision unit provides accurate lateral distance measurement. The system fuses data from both sensors to achieve both reliability and precision simultaneously.
2Reliability
If radar-based detection units are used to detect targets, then detection reliability in low visibility conditions is improved, but lateral distance detection accuracy deteriorates
Solution Approach 1:
The vision-based detection unit acts as an intermediary to correct the lateral distance measurement errors of the radar system. The radar provides target detection and range information, while the vision system refines the lateral position data, combining the strengths of both sensors.
3Device complexity
If single detection system is used, then device complexity is reduced, but false detection rate increases
Solution Approach 1:
The system uses feedback from multiple detection units to verify target detections. When both radar and vision systems detect the same target, the system gains confidence in the detection. This cross-verification mechanism reduces false detections while maintaining system manageability.
4Measurement precision
If vision-based detection is used in cluttered environments, then target identification accuracy is improved, but detection system fails in highly cluttered environments
Solution Approach 1:
The system dynamically switches between radar-based and vision-based detection modes depending on environmental conditions. In highly cluttered environments where vision-based detection fails, the system relies on radar-based detection which maintains functionality, ensuring continuous operation across varying conditions.
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
This approach reduces dependency on vision-based detection in low visibility conditions, improves accuracy by overcoming inaccuracy in lateral distance determination by radar, and minimizes false brake conditions, enabling stable target selection and control with minimal overlap between the host vehicle and target.
Implementation Method 1
a radar based detection unit, comprising one or more radar sensors, for detecting one or more targets around the host vehicle
Implementation Method 2
a vision based detection unit, comprising one or more image sensors, for detecting one or more targets in the field of view of the host vehicle
Data Source
AI summary
An aspect of the present disclosure relates to a system implemented in a host vehicle, comprising: a radar based detection unit, comprising one or more radar sensors, for detecting one or more targets around the host vehicle; a vision based detection unit, comprising one or more image sensors, for detecting one or more targets in the field of view of the host vehicle; and a processing unit to: receive information corresponding to detected one or more targets from each of the radar based detection unit and the vision based detection unit; match each of the one or more targets detected by the radar based detection unit with one or more targets detected by vision based detection unit to identify a target as a matched target; categorize the matched target as a locked target; and track the locked target using information received from the radar based detection unit.