Multi-Sensor Fusion for Accurate Vehicle Sensor Cleanliness Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current intelligent driving systems face challenges in accurately determining the clean status of sensors, such as cameras and lidars, which are crucial for autonomous driving, as they are often affected by environmental factors like rain or snow, leading to false obstacle detection or missed targets, compromising safety.
Innovation Solution
A sensor detection method that involves data collection from multiple sensors, feature extraction, data fusion, and inference to determine the clean status of each sensor, using encoders and decoders, either rule-based or neural networks, to improve accuracy and trigger appropriate cleaning actions or alert systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If simple sensors (e.g., rainfall sensor) are used to determine sensor cleanliness, then the detection system remains simple, but the accuracy of determining actual sensor clean status deteriorates
Solution Approach 1:
The patent combines data from multiple sensors (camera, lidar, millimeter-wave radar, ultrasonic sensor) to comprehensively determine sensor clean status. By merging information from different sensing modalities, the system achieves accurate detection of whether sensors are blocked by dust, rain, snow, or other contaminants, resolving the contradiction between system simplicity and detection accuracy.
Solution Approach 2:
The detection system is designed to handle multiple types of contamination (dust, rain, snow, ice) using a unified multi-sensor approach. The same sensor fusion framework universally detects various blocking conditions across different sensor types, providing versatile and accurate clean status determination without requiring separate specialized detection mechanisms for each contamination type.
2Measurement precision
If multiple sensors are used for data fusion, then the accuracy of clean status detection improves, but the device complexity increases
Solution Approach 1:
The patent segments the detection task by assigning specific functions to different sensor types. Each sensor (camera for visual obstruction, lidar for point cloud blockage, millimeter-wave radar for electromagnetic wave interference, ultrasonic sensor for acoustic wave obstruction) detects specific aspects of contamination. This segmentation allows the complex detection problem to be divided into manageable sub-tasks, improving accuracy while organizing system complexity in a structured manner.
Solution Approach 2:
The patent introduces a sensor fusion module as an intermediary that processes and integrates data from multiple sensors. This mediator combines information from camera, lidar, millimeter-wave radar, and ultrasonic sensor to produce a comprehensive clean status determination. The intermediary handles the complexity of multi-sensor integration, allowing individual sensors to remain relatively simple while achieving high overall detection accuracy.
Data Source
AI summary
A sensor detection method, a sensor detection apparatus, and a vehicle are provided. The method may be applied to the sensor detection apparatus. The method includes: a sensor detection apparatus obtains data collected by a plurality of sensors; the sensor detection apparatus extracts a feature from data collected by each of the plurality of sensors, to obtain a plurality of pieces of feature data; the sensor detection apparatus fuses the plurality of pieces of feature data to obtain fused data; and the sensor detection apparatus performs inference on the fused data to obtain clean status information of each of the plurality of sensors. According to embodiments of this application, accuracy of detecting a clean status of a sensor can be improved.