Lane Assist System Precipitation Detection via Bayesian Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in maintaining safe and efficient operation during precipitation and other adverse weather conditions, as sensors can be obscured, leading to inaccurate data and potential lane deviation.
Innovation Solution
A method using Bayesian inference to determine the probability of precipitation based on vehicle wiper status, vehicle jerk, and vehicle lateral offset, which enables the disabling and re-enabling of the lane assist system to ensure safe operation by analyzing sensor data and confidence levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If the lane assist system operates during precipitation, then the vehicle can maintain autonomous lane keeping, but sensor accuracy deteriorates leading to lane deviation
Solution Approach 1:
The system performs preliminary detection of precipitation conditions using wiper status, vehicle jerk, and lateral offset sensors before lane deviation occurs. By predicting precipitation probability in advance, the system can prepare to disable lane assist proactively, preventing the harmful effect of sensor obscuration on measurement precision while maintaining automation during clear conditions.
2Measurement precision
If the lane assist system is disabled during precipitation, then sensor accuracy is preserved, but vehicle productivity decreases due to manual operation requirement
Solution Approach 1:
Instead of completely disabling lane assist during all adverse conditions, the system applies partial action by disabling it only when precipitation probability exceeds a specific threshold. This selective approach maintains measurement precision when needed while preserving productivity during mild conditions where sensors remain functional, avoiding unnecessary manual operation.
Solution Approach 2:
The system continuously monitors wiper status, vehicle jerk, and lateral offset to provide feedback on precipitation conditions. This real-time feedback loop allows dynamic adjustment of lane assist operation, disabling only when sensor accuracy deteriorates below acceptable levels while maintaining automation when conditions permit, thus balancing measurement precision and productivity.
3Reliability
If multiple sensors are used to detect precipitation, then detection reliability improves, but device complexity increases
Solution Approach 1:
The system uses existing vehicle components (wiper motor status, vehicle dynamics sensors for jerk and lateral offset) that already serve other functions to also detect precipitation conditions. This self-service approach improves detection reliability by utilizing multiple existing sensors without adding dedicated precipitation detection hardware, thus avoiding increased device complexity.
Data Source
AI summary
A computing system can determine a probability of precipitation based on Bayesian inference conditioned on probabilities associated with vehicle wiper status, vehicle jerk, vehicle sway, and vehicle lateral offset. The computing system can disable a vehicle lane assist system based on the probability of precipitation and operate a vehicle with the disabled vehicle lane assist system.