External-Environment Sensor Diagnosis Using Dynamic Reference Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor failure detection systems rely on pre-registered map information, leading to erroneous diagnoses when static objects are moved and fail to function in areas with few static objects, limiting diagnosis to urban areas.
Innovation Solution
A sensing performance evaluation and diagnosis system that acquires sensing data from external-environment sensors and compares it with reference output from other sensors, using a reference value calculation unit to evaluate performance deterioration without relying on pre-registered map information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pre-registered map information is used as reference for sensor diagnosis, then diagnosis can be performed in urban areas with many static objects, but diagnosis becomes inaccurate when static objects are moved and cannot be performed in areas with few static objects
Solution Approach 1:
The patent creates a dynamic reference output by copying the sensing data from the diagnosis target sensor at a previous time point, rather than relying on static pre-registered map information. This copied historical data serves as the reference for comparison, enabling accurate diagnosis regardless of whether static objects are present or have moved.
Solution Approach 2:
The patent transforms the static reference information (pre-registered map data) into dynamic reference information by using time-series sensing data from the sensor itself. The reference output is continuously updated based on recent historical data, making it adaptable to changing environments and object positions.
2Measurement precision
If static objects registered on map are used as reference, then diagnosis can be performed using known position information, but the system fails when these objects are moved due to road construction or in areas without such objects
Solution Approach 1:
The sensor diagnoses itself by comparing its current output with its own previous outputs, rather than relying on external static reference objects. The sensing data from the diagnosis target sensor serves dual purposes: as both the object to be measured and as the reference standard for self-diagnosis.
Solution Approach 2:
The system stores sensing data from previous time points as reference data in advance, so that when diagnosis is needed, the comparison can be immediately performed without requiring external reference objects. This preliminary storage of historical data enables rapid and reliable diagnosis.
Data Source
AI summary
An object of the present invention is to provide a sensing performance evaluation and diagnosis system capable of detecting a failure sign of an external-environment recognition sensor as a diagnosis target without relying on reference information registered in advance on a map, by comparing an output of the external-environment recognition sensor as the diagnosis target with a reference output of the external-environment recognition sensor. The sensing performance evaluation and diagnosis system includes a sensing data acquisition unit that acquires sensing data around an own vehicle from an observation value of an external-environment sensor as a diagnosis target, which is mounted on the own vehicle, a surrounding information acquisition unit that acquires surrounding information data around the own vehicle from an observation value of a reference external-environment sensor, a reference value calculation unit that generates reference value data based on a recognition history of relative information to a reference object, which is included in the surrounding information data, and an evaluation unit that evaluates whether or not performance of the external-environment sensor as the diagnosis target is deteriorated or evaluates a degree of performance deterioration, by comparing the sensing data related to the reference object to a threshold value set for the reference value data.