Radar-Camera Fusion for Object Detection with Data Extrapolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High manufacturing costs of object detection devices due to the need for high-performance cameras for accurate object detection.
Innovation Solution
An object detection apparatus that combines a radar for emitting and receiving electromagnetic waves with a low-performance camera for image capture, using a fusion processor to generate and extrapolate position and velocity data even when radar data is lost, allowing for accurate detection at lower costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a high-performance camera is used to accurately detect object position data, then detection accuracy is improved, but manufacturing cost increases
Solution Approach 1:
The patent uses a low-performance, inexpensive camera instead of a high-performance camera to detect object position data. The system compensates for the lower camera quality by using radar data and data from previous frames to generate accurate object detection results, thereby achieving good detection accuracy with a cheap camera.
Solution Approach 2:
The patent combines data from multiple sources including radar position data, radar velocity data, camera position data, and camera velocity data from previous frames to generate comprehensive object detection results. This fusion of multiple data sources compensates for the limitations of using a low-performance camera.
2Reliability
If radar data is lost, then detection reliability deteriorates, but system complexity increases when implementing data extrapolation
Solution Approach 1:
The system stores position data and velocity data from previous frames in advance. When radar data is lost, the fusion processor can immediately use the pre-stored data to generate extrapolated detection results without needing to perform complex real-time calculations, thereby maintaining detection reliability with minimal additional complexity.
Solution Approach 2:
The fusion processor creates a copy of the previous frame's detection results and uses this copied data as a basis for generating current frame detection results when radar data is unavailable. This copying approach simplifies the data processing compared to performing full real-time extrapolation calculations.
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
Enables accurate object detection at reduced manufacturing costs by utilizing low-performance cameras and extrapolating data from previous frames when radar data is unavailable, improving detection performance and reducing hardware and software loads.
Implementation Method 1
a radar to emit an electromagnetic wave toward an object and receive a reflected signal from the object
Data Source
AI summary
An object detection apparatus includes a signal processor that computes radar position data and radar velocity data on the basis of a reflected signal received by a radar; an image processor that computes camera velocity data indicating a velocity of the object on the basis of image information obtained by a camera; and a fusion processor that outputs, to a vehicle control device, the radar position data and the radar velocity data of a first frame as first detected position data and first detected velocity data. When the radar position data and the radar velocity data of a second frame following the first frame are lost, the fusion processor generates, on the basis of the first detected position data and the camera velocity data of the second frame, and outputs, to the external device, second detected position data and second detected velocity data for the second frame.