Autonomous Vehicle Sensor Cross-Validation Using Object State Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face navigation challenges due to sensor failures or inaccuracies, which can lead to decreased confidence and precision in detecting objects in their environment.
Innovation Solution
The system employs cross-validation between different types of sensors, such as cameras and lasers, by comparing object detection parameters and state information to determine if one sensor is functioning correctly, using a processor to select reference sensors and perform cross-validation based on time intervals or failure thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple sensors are used to detect objects in the driving environment, then the reliability of object detection is improved, but the device complexity increases
Solution Approach 1:
The system implements cross-validation by comparing sensor readings from multiple sensors against each other and against expected values from the detailed map. When a sensor reading deviates from the expected range, the system identifies it as potentially faulty and adjusts reliance on that sensor, creating a feedback mechanism that continuously monitors and adjusts sensor reliability.
Solution Approach 2:
Multiple sensors serve dual functions: they detect objects in the driving environment and simultaneously validate each other's readings. The same sensor array used for primary detection also performs cross-validation, eliminating the need for separate validation hardware and reducing overall system complexity.
2Measurement precision
If sensor cross-validation is performed continuously, then the measurement precision of sensor data is improved, but the loss of time increases
Solution Approach 1:
Instead of continuous cross-validation, the system performs validation at periodic intervals or triggered by specific events such as detected deviations or changes in driving conditions. This periodic approach maintains measurement precision while significantly reducing the time loss associated with constant validation processing.
Solution Approach 2:
The system performs cross-validation selectively on specific sensors or specific parameters based on pre-established criteria rather than validating all sensor data uniformly. This partial validation approach focuses computational resources on the most critical or suspicious readings, improving precision where needed while minimizing overall processing time.
3Measurement precision
If deviation thresholds are set low for strict sensor validation, then the measurement precision is improved, but the reliability decreases due to false failures
Solution Approach 1:
The system dynamically adjusts validation thresholds based on environmental conditions, sensor type, and driving context rather than using fixed thresholds. This allows the system to maintain high measurement precision in controlled conditions while being more tolerant of normal variations in challenging environments, reducing false failure detections.
Solution Approach 2:
Different validation criteria and thresholds are applied to different sensors and different types of readings based on their specific characteristics and reliability profiles. Critical safety-related parameters use stricter validation, while less critical parameters use more lenient thresholds, optimizing both precision and reliability for each specific case.
Data Source
AI summary
Methods and systems are disclosed for cross-validating a second sensor with a first sensor. Cross-validating the second sensor may include obtaining sensor readings from the first sensor and comparing the sensor readings from the first sensor with sensor readings obtained from the second sensor. In particular, the comparison of the sensor readings may include comparing state information about a vehicle detected by the first sensor and the second sensor. In addition, comparing the sensor readings may include obtaining a first image from the first sensor, obtaining a second image from the second sensor, and then comparing various characteristics of the images. One characteristic that may be compared are object labels applied to the vehicle detected by the first and second sensor. The first and second sensors may be different types of sensors.