AV Road Anomaly Detection via Sensor Log Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles (AVs) face issues with misalignment and potential damage due to road anomalies such as potholes, which can cause them to prematurely shut down or transition to safety mode, leading to unnecessary autonomy shutdowns and potentially dangerous conditions.
Innovation Solution
A network computing system that coordinates AVs by analyzing sensor log data against road anomaly signatures, classifying anomalies, and generating resolution responses to preemptively adjust routes, speed, or switch to teleassistance mode to avoid or mitigate the impact of road anomalies, using updated localization maps and individual vehicle characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the AV control system uses acceleration thresholds to detect road anomalies, then the system can identify potential road hazards, but the AV will prematurely transition to safety mode causing unnecessary autonomy shutdowns
Solution Approach 1:
The system performs preliminary classification of acceleration events by comparing sensor data against stored road anomaly signatures before triggering autonomy shutdown. This preliminary action distinguishes between benign road anomalies and genuine hazards, allowing the AV to maintain autonomous operation while still detecting potential road hazards accurately
Solution Approach 2:
The system uses feedback from sensor log data and acceleration patterns to continuously refine the distinction between benign and harmful acceleration events. By analyzing historical data and comparing against known road anomaly signatures, the system learns to differentiate normal road conditions from genuine hazards, reducing false positives while maintaining detection accuracy
2Object-affected harmful factors
If the AV transitions to safety mode upon detecting road anomalies, then the system prioritizes vehicle safety, but operational disruptions and unnecessary manual intervention increase
Solution Approach 1:
The system applies different response strategies based on the specific type and severity of road anomaly detected. Benign anomalies like minor potholes or expansion joints trigger no autonomy shutdown, while only severe hazards trigger safety mode transitions. This localized response quality maintains productivity by avoiding unnecessary manual intervention while still preventing vehicle damage from genuine hazards
Solution Approach 2:
The system changes the acceleration threshold parameters dynamically based on the classified type of road anomaly. Different anomaly types have different safe acceleration thresholds, allowing the system to tolerate higher accelerations for benign conditions while maintaining strict protection for hazardous conditions, thereby balancing safety with operational efficiency
Data Source
AI summary
A computing system can receive sensor log data from one or more first autonomous vehicles (AVs) operating along one or more routes. The system can analyze the sensor log data to identify a road anomaly along the one or more routes, and generate an updated localization map comprising a label that indicates the road anomaly. The system may then transmit the updated localization map to one or more second AVs to enable the one or more second AVs to respond to the road anomaly.