Fusion Sensor Distance Correction via Radar-Vision Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating distances to objects, such as pedestrians, using image sensors and radar sensors suffer from significant errors and difficulties in distinguishing and recognizing pedestrians, necessitating a more accurate approach.
Innovation Solution
A device and method utilizing a fusion sensor that combines data from a radar or lidar sensor and an image sensor to correct distance estimates by assigning vision tracks to clusters based on closest radar tracks, determining a correction ratio, and applying it to all vision tracks within the cluster, thereby improving distance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If distance calculation is performed using image sensor only, then the system is simple, but the distance calculation has large error with actual distance
Solution Approach 1:
The patent combines radar sensor and image sensor into a fusion sensor system. The radar sensor detects positions of multiple objects while the image sensor captures front images. By merging the detection results of both sensors and performing cluster-based association, the system achieves accurate distance measurement that overcomes the limitations of using either sensor alone.
2Area of stationary object
If distance calculation is performed using radar sensor only, then the detection range is extended, but the system cannot distinguish and recognize pedestrians
Solution Approach 1:
The patent merges radar sensor detection results with image sensor recognition capabilities. The radar sensor provides extended detection range by detecting object positions, while the image sensor provides pedestrian recognition information through image capture. The fusion sensor system associates results from both sensors to achieve both extended detection range and accurate pedestrian identification.
3Measurement precision
If fusion sensor is used to combine radar and image sensor, then distance accuracy is improved, but the device complexity increases
Solution Approach 1:
The patent segments the fusion sensor system into distinct functional modules: radar sensor for position detection, image sensor for image capture, and a controller for processing. The controller further segments the processing into track recognition, cluster assignment, and distance correction stages. This segmentation manages complexity by organizing functions into manageable, independent components.
Solution Approach 2:
The patent introduces a controller as an intermediary that mediates between the radar sensor and image sensor. The controller receives detection results from both sensors, performs cluster-based association to match radar tracks with vision tracks, and calculates corrected distances. This intermediary manages the complexity of sensor fusion by providing a centralized processing layer that coordinates data from multiple sources.
Data Source
AI summary
A device for estimating a distance based on object detection and a method thereof are provided. The device for estimating a distance based on object detection according to an embodiment of the present disclosure includes a fusion sensor including a first sensor configured to detect positions of a plurality of objects in front of a host vehicle and a second sensor configured to capture a front image of the host vehicle, and a controller communicatively connected to the fusion sensor and configured to recognize all radar tracks corresponding to distances detected by the first sensor and all vision tracks corresponding to distances detected by the second sensor, assign adjacent vision tracks for each of the radar tracks to one cluster, and correct distances of all the vision tracks assigned to the corresponding cluster based on the closest vision track from the radar track for each cluster.