Vehicle Trailer Vision Using DNN Detection and Auto Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle vision systems struggle to accurately detect and classify trailers of various types, orientations, and environmental conditions, which hinders autonomous alignment and navigation of vehicles with trailers.
Innovation Solution
A vehicle vision system utilizing one or more cameras and a deep neural network (DNN) to process image data, detect the presence of a trailer, determine its position and category, and autonomously align the vehicle with the trailer.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing methods are used to detect trailers, then the system complexity is low, but the detection accuracy and classification precision are insufficient
Solution Approach 1:
The patent replaces traditional mechanical image processing methods with a deep neural network (DNN) based vision system. The DNN automatically learns features from raw image data, substituting manual feature extraction and processing algorithms with an intelligent system that achieves superior detection accuracy and classification precision for trailers in various conditions.
Solution Approach 2:
The patent transforms the detection approach by changing from fixed threshold-based parameters to adaptive parameters learned by the DNN. The neural network dynamically adjusts detection parameters based on environmental conditions, trailer orientations, and image quality, enabling accurate detection across diverse scenarios without manual parameter tuning.
2Adaptability or versatility
If the system attempts to detect all types of trailers in all environmental conditions, then the versatility is improved, but the reliability of detection decreases due to complexity
Solution Approach 1:
The patent segments the detection task into multiple specialized DNN models or detection stages, each optimized for specific trailer types or environmental conditions. This modular approach allows the system to maintain high reliability for each segment while achieving comprehensive versatility through the combination of multiple specialized detectors.
Solution Approach 2:
The system performs preliminary classification of detected objects to identify potential trailers before applying full detection algorithms. This preliminary action filters out non-trailer objects early in the processing pipeline, reducing false positives and improving overall detection reliability while maintaining the ability to detect diverse trailer types.
3Extent of automation
If manual alignment and navigation to trailer is used, then the automation level is low, but the ease of operation is maintained
Solution Approach 1:
The vision system performs autonomous alignment and navigation to the trailer without requiring manual driver intervention. The DNN processes camera images, calculates the optimal path to the trailer, and controls the vehicle's steering and movement automatically, enabling the system to service itself in completing the hitching task.
Solution Approach 2:
The system continuously captures images during the alignment process, processes them through the DNN to determine current position relative to the trailer, and adjusts the vehicle's movement in real-time based on this feedback. This closed-loop control enables precise autonomous alignment while simplifying the operator's role to monitoring and oversight.
Data Source
AI summary
A vehicular trailer assist system includes a camera disposed at a rear portion of a vehicle and viewing at least rearward of the vehicle, and an electronic control unit (ECU), which includes an image processor for processing image data captured by the camera. The ECU executes a deep neural network (DNN) to determine presence of a trailer in image data captured by the camera. The DNN, responsive to processing of image data captured by the camera, determines a position of a trailer in the captured image data. The DNN, responsive to determining the position of the trailer, classifies the trailer into a trailer category. The ECU, responsive to the DNN determining the position of the trailer and based on the classification, generates an output to autonomously control the vehicle to align the vehicle with the trailer and navigate the vehicle toward the trailer.