Autonomous Driving Abnormality Detection via Reference Travel Deviation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous driving systems face challenges in accurately detecting abnormalities, as they rely on recognition performance, which can lead to failure in recognizing objects or erroneous detection, making it difficult to determine the presence of abnormalities.
Innovation Solution
An abnormality detection system that compares the travel plan of a target vehicle with the travel record of a reference vehicle, calculating deviations at specific positions to identify abnormal positions, thereby detecting system abnormalities independently of recognition performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the autonomous driving system uses recognition performance to determine abnormality, then the abnormality detection can be performed using existing sensors, but the detection accuracy deteriorates when objects are not recognized or erroneous recognition occurs
Solution Approach 1:
The patent introduces reference vehicle travel information as an intermediary to compare against the target vehicle's actual travel behavior. Instead of directly relying on sensor recognition accuracy, the system uses the reference travel information (stored in storage device 122) as a mediator to identify deviations in the target vehicle's travel plan, thereby detecting abnormalities without being constrained by recognition system performance
Solution Approach 2:
The patent creates a copy of normal travel behavior through reference vehicle data. By storing and comparing against reference travel information from vehicles that have successfully completed the same route, the system establishes a baseline for normal operation. This copying approach allows abnormality detection based on behavioral patterns rather than direct sensor recognition, bypassing the recognition accuracy limitation
2Device complexity
If the system continuously monitors recognition stability to detect abnormalities, then the detection method is simple, but it cannot identify abnormalities when recognition fails or produces continuous errors
Solution Approach 1:
Instead of monitoring whether the recognition system produces stable results, the patent inverts the approach by monitoring whether the actual travel behavior deviates from the planned travel behavior (which was generated based on recognition). The abnormality detection device 120 compares the travel plan with actual travel information, detecting abnormalities when the vehicle fails to execute the planned route, regardless of whether the recognition system reports stable but potentially erroneous data
Data Source
AI summary
An abnormality detection system is applied to an autonomous driving system of a target vehicle. The abnormality detection system stores travel plan information and reference travel information. The travel plan information and the reference travel information indicate a travel plan of the target vehicle and a travel record of a reference vehicle different from the target vehicle, respectively. The abnormality detection system calculates a deviation between the travel plan of the target vehicle and the travel record of the reference vehicle for each determination position based on the travel plan information and the reference travel information. The abnormality detection system extracts the determination position at which the deviation exceeds a threshold as an abnormal position related to an abnormality of the autonomous driving system. The abnormality detection system enables to appropriately detect the abnormality of the autonomous driving system independently of the recognition performance of the autonomous driving system.