Road Object Localization Verification by Image Reprojection Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately localizing objects, such as stop signs, due to varying estimates from different localization techniques, leading to potential incorrect reactions and unsafe navigation.
Innovation Solution
A method and system that utilize multiple localization techniques (object size, depth map, and roadgraph-based) to identify and verify the location of objects by calculating error scores based on reprojection errors, selecting the most accurate estimate for controlling the vehicle's autonomous driving mode.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple localization techniques are used to determine object locations, then the reliability of object localization is improved, but the device complexity increases
Solution Approach 1:
The patent combines multiple localization techniques (object size-based localization, depth map-based localization, and roadgraph-based localization) into a unified system that processes results from all techniques and selects the most accurate estimate through error score evaluation and reprojection analysis
Solution Approach 2:
The system uses reprojection error calculation as a feedback mechanism to evaluate the accuracy of each localization technique's estimate. By projecting estimated object locations back into image space and comparing with actual detections, the system generates error scores that guide the selection of the most reliable localization result
2Reliability
If multiple localization techniques are used to determine object locations, then the reliability of object localization is improved, but the computational resources required increase
Solution Approach 1:
The system applies multiple localization techniques but only processes a subset of them to full completion. It calculates error scores for each technique and selectively performs detailed reprojection analysis only for estimates with lower error scores, avoiding full processing of all techniques when unnecessary
3Measurement precision
If reprojection error calculation is performed for each estimated location, then the measurement precision of object location is improved, but the computational time required increases
Solution Approach 1:
The system calculates reprojection errors for each estimated location but uses these calculations selectively. It first computes error scores from multiple localization techniques, then performs detailed reprojection analysis primarily on estimates with lower error scores, rather than exhaustively processing all estimates with equal depth
Data Source
AI summary
Aspects of the disclosure relate to verifying the location of an object of a particular type. For instance, a plurality of images of an environment of the vehicle may be received. Associated objects of the particular type may be identified in ones of the plurality of images. A plurality of estimated locations may be determined for the object using a plurality of different localization techniques. For each image of the ones of the plurality of images, determine a reprojection error for each of the plurality of estimated locations. For each of the plurality of estimated locations, an error score is determined based on the reprojection errors. An estimated location may be selected from the plurality of estimated locations based on the determined error score. This selected location may be used to control a vehicle in an autonomous driving mode.