Lidar Point Cloud Anomaly Detection via Remote Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in effectively detecting and managing lidar point cloud anomalies, which can impact their navigation and safety, particularly in situations like construction sites or emergency scenarios, where existing systems may fail to accurately identify and respond to such anomalies in real-time.
Innovation Solution
A system comprising an autonomous vehicle computer module and a remote computer module that collaboratively analyze lidar data to detect potential anomalies, confirm them, and take appropriate actions, such as rerouting, by comparing current lidar point clouds with prior data and receiving remote confirmation and assistance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the autonomous vehicle uses onboard processor to analyze lidar point cloud data for anomaly detection, then the response speed is improved, but the measurement precision and reliability deteriorate due to limited computational resources and potential false positives
Solution Approach 1:
A remote computing system acts as an intermediary between the autonomous vehicle's onboard processor and the anomaly detection task. The onboard processor identifies potential anomalies and sends them to the remote system for verification, which then returns confirmation or correction. This distributed architecture allows the vehicle to maintain fast local response while leveraging remote computational power for high-precision anomaly verification, resolving the contradiction between speed and accuracy.
2Reliability
If the autonomous vehicle implements comprehensive lidar point cloud anomaly detection, then the safety is improved, but the device complexity increases due to additional sensors and processing requirements
Solution Approach 1:
The system implements self-service by using the existing lidar sensors and onboard processors already present in autonomous vehicles for anomaly detection, rather than adding dedicated anomaly detection hardware. The remote computing system provides additional processing capabilities without requiring physical installation on the vehicle. This approach improves safety through comprehensive anomaly detection while avoiding increased device complexity by leveraging existing components.
3Measurement precision
If the autonomous vehicle continuously monitors and compares lidar point cloud data with prior information, then the anomaly detection capability is improved, but the energy consumption increases
Solution Approach 1:
The system applies partial action by performing complete anomaly detection only when potential anomalies are detected by the onboard processor, rather than continuously analyzing all lidar data. The remote computing system is engaged selectively for verification of suspected anomalies, and the comparison with prior point cloud information is performed only when needed. This approach maintains high detection capability while reducing overall energy consumption by avoiding unnecessary full-scale analysis during normal operation.
Data Source
AI summary
Systems and method are provided for controlling an autonomous vehicle. In one embodiment, a method for controlling an autonomous vehicle comprises obtaining lidar data from one or more lidar sensors disposed on the autonomous vehicle during operation of the autonomous vehicle, generating a lidar point cloud using the lidar data, making an initial determination, via a processor onboard the autonomous vehicle, of a possible lidar point cloud anomaly based on a comparison of the lidar point cloud with prior lidar point cloud information stored in memory, receiving a notification from a remote module as to whether the possible lidar point cloud anomaly is a confirmed lidar point cloud anomaly, and taking one or more vehicle actions when there is a confirmed lidar point cloud anomaly.