Vehicle Sensor Calibration Using Infrastructure Signal Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous and semi-autonomous vehicle systems face challenges in accurately identifying traffic signs and signals, especially in areas without smart infrastructure, as they rely solely on onboard sensors which can be obstructed or lack necessary data for optimal operation.
Innovation Solution
A vehicle sensor training system that combines data from onboard sensors with infrastructure signals to calibrate and train the vehicle sensor, using a computing device to receive and compile dynamics data and real-time object identification, enabling the vehicle to identify traffic signs and signals even in the absence of vehicle-to-infrastructure communication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the vehicle relies solely on onboard sensors to identify traffic signs and signals, then the system complexity is reduced, but the detection accuracy and reliability deteriorate due to obstructions and lack of data
Solution Approach 1:
The patent combines onboard vehicle sensors with infrastructure-based signals (V2I communication) to create a hybrid detection system. The computing device receives and processes both sensor data from the vehicle and signal data from infrastructure, merging multiple data sources to improve detection accuracy and reliability while maintaining manageable system complexity through integrated processing.
2Loss of information
If smart infrastructure systems are implemented at each intersection to provide traffic information, then the information accuracy is improved, but the implementation cost and time increase significantly
Solution Approach 1:
The system performs preliminary calibration by collecting and processing infrastructure signal data during normal operation to build a database of traffic signs and signals. This preliminary action allows the vehicle sensor to be trained in advance, reducing the need for expensive smart infrastructure deployment at every intersection while still achieving high information accuracy through pre-collected data.
Solution Approach 2:
The vehicle system calibrates and trains its own sensors using infrastructure signal data when available, making the system self-sufficient. The computing device automatically processes infrastructure information to improve sensor accuracy without requiring external intervention or expensive infrastructure modifications, allowing the system to serve itself through autonomous calibration.
3Measurement precision
If the vehicle sensor is trained using only onboard sensor data, then the system is simpler to operate, but the sensor cannot accurately identify traffic signs in challenging conditions such as low visibility or at high speeds
Solution Approach 1:
The system uses infrastructure signal data as feedback to calibrate and validate sensor readings. The computing device compares sensor-identified traffic signs with infrastructure-provided signal information, using this feedback loop to continuously improve sensor accuracy. This feedback mechanism enables the sensor to achieve high measurement precision in challenging conditions while the automated feedback process maintains operational simplicity.
Data Source
AI summary
A vehicle sensor training system includes a first computing device configured to receive dynamics data from a host vehicle. A receiving device is configured to receive real time object identification data from a reference target. A second computing device is configured to compile the dynamics data from the first computing device and the real-time object identification data from the receiving device. The second computing device is configured to train a vehicle sensor with the real-time object identification data.