Trailer Detection Using Convolution Filters for Reverse Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to accurately and efficiently determine the position of a trailer relative to a vehicle, especially when reversing, due to the trailer's pivoting motion, which can lead to contact with the vehicle's rear portion.
Innovation Solution
A method using real-time image processing on a vehicle's controller, involving convolution of feature patches with filters based on known trailer data, and a voting process to identify the trailer's location, reducing the search area by comparing consecutive images for stationary objects, and utilizing a database of rear-view images to optimize filter development.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If real-time image processing with convolution and voting process is implemented on the vehicle's controller, then the accuracy of trailer position determination is improved, but the computing power requirement increases beyond the capability of the vehicle's controller
Solution Approach 1:
The system performs preliminary actions by pre-processing images of known trailers offline to generate filter templates (H) and storing them in memory before real-time operation. This preprocessing step extracts key features and creates convolution kernels that can be efficiently applied during real-time detection, reducing the computational burden on the vehicle's controller while maintaining high accuracy in trailer position determination
Solution Approach 2:
The image processing is segmented into distinct stages: offline pre-processing to generate filters, real-time feature extraction from captured images, convolution operation with pre-computed filters, and voting process for position determination. This segmentation allows computationally intensive operations to be performed offline while the vehicle controller only handles lighter real-time operations, resolving the contradiction between accuracy and computing power constraints
2Reliability
If the driver manually monitors the trailer angle continuously, then the trailer alignment can be controlled, but the driver's attention is consumed and the operation complexity increases
Solution Approach 1:
The system implements self-service by automatically detecting trailer position and providing guidance information without requiring continuous manual monitoring by the driver. The image processing system, convolution algorithms, and voting process work autonomously to determine trailer location and alignment, freeing the driver's attention while maintaining reliable trailer control through automated assistance
Data Source
AI summary
A method for determining a location of a trailer in an image includes obtaining at least one real-time image from a vehicle. The at least one real-time image is processed with a controller on the vehicle to obtain a feature patch describing at least one real-time image. A convolution is performed of the feature patch and each filter from a set of filters with the filter being based on data representative of known trailers. A location of a trailer is determined in the at least one real-time image based on the convolution between the feature patch and each filter from the set of filters.