Reference Image Obstacle Detection for Autonomous Road Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in detecting road obstacles and changes effectively, particularly when background objects are static and need to differentiate between moving objects and static road features for safe navigation.
Innovation Solution
A computing device receives images from a camera and compares them to a reference image to identify differences, determining the level of confidence in these differences, and modifies the vehicle's control strategy based on this information to adjust driving behavior and control the vehicle accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vehicle uses image comparison to detect road obstacles, then the detection accuracy is improved, but the computational complexity and processing time increase
Solution Approach 1:
The system pre-processes images to identify and mask static background objects before comparing current images with reference images. By removing static elements in advance, the comparison process focuses only on potential obstacles, significantly reducing computational complexity while maintaining high detection accuracy
Solution Approach 2:
The image processing is divided into distinct segments: static object identification, static object masking, and obstacle detection. This segmentation allows each processing stage to be optimized independently, reducing overall computational burden while preserving detection precision
2Reliability
If the vehicle continuously monitors road changes using image comparison, then the safety is improved, but the energy consumption increases
Solution Approach 1:
Instead of continuous real-time comparison, the system uses periodic sampling with reference images captured at intervals. This periodic monitoring maintains safety by detecting obstacles when they appear or change, while significantly reducing energy consumption compared to continuous frame-by-frame analysis
Solution Approach 2:
The system extracts only the essential comparison elements (differences between current and reference images) rather than processing entire continuous video streams. This extraction approach maintains safety monitoring while reducing the energy burden of constant full-image processing
3Measurement precision
If the vehicle differentiates between moving objects and static road features, then the navigation accuracy is improved, but the difficulty of detecting and measuring increases
Solution Approach 1:
The system performs preliminary classification by identifying static road features and masking them before obstacle detection. This pre-separation simplifies the subsequent detection task, making it easier to distinguish moving obstacles from static background while maintaining high navigation accuracy
Solution Approach 2:
The masking process acts as an intermediary that separates static and moving elements. By creating a masked version of the reference image, the system facilitates easier comparison and classification, reducing the difficulty of distinguishing between static features and moving obstacles
Data Source
AI summary
Methods and systems for use of a reference image to detect a road obstacle are described. A computing device configured to control a vehicle, may be configured to receive, from an image-capture device, an image of a road on which the vehicle is travelling. The computing device may be configured to compare the image to a reference image; and identify a difference between the image and the reference image. Further, the computing device may be configured to determine a level of confidence for identification of the difference. Based on the difference and the level of confidence, the computing device may be configured to modify a control strategy associated with a driving behavior of the vehicle; and control the vehicle based on the modified control strategy.