Camera-Based Yaw Rate Correction for Autonomous Vehicle Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating a vehicle's yaw rate in autonomous driving face significant uncertainties due to thermal drift in inertial sensors and poor GNSS signal reception, particularly in urban areas, which can lead to inaccurate control signals and potential failures during emergency maneuvers.
Innovation Solution
A method that combines sensor data from inertial measurement units, GNSS sensors, wheel velocity sensors, and steering angle sensors with camera-based optical detection to correct yaw rate estimates, using a camera-based yaw rate value to offset and improve the accuracy of inertial and GNSS-based estimates, ensuring reliable control signals even under sensor failures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor fusion of inertial measurement unit and GNSS is used to estimate yaw rate, then the method provides comprehensive vehicle motion information, but the measurement precision deteriorates due to thermal drift of inertial sensors and poor GNSS signal reception in urban areas
Solution Approach 1:
The patent introduces camera-based optical flow detection as an intermediary measurement system to bridge the gap between inertial sensors and ground truth yaw rate. The camera captures visual information from which optical flow fields are computed, serving as a mediator to correct drift accumulation in inertial measurement units without requiring direct GNSS signal reception.
Solution Approach 2:
The system implements feedback by continuously comparing inertial measurement unit data with camera-based optical flow measurements. The computed yaw rate from optical flow serves as feedback to correct deviations in the inertial sensor estimates, creating a closed-loop system that maintains measurement precision over time despite thermal drift and signal reception issues.
2Quantity of substance
If multiple information sources including inertial sensors and GNSS are fused to increase signal quality, then the quantity of information increases, but the measurement precision worsens due to uncertainties from thermal drift and reception difficulties
Solution Approach 1:
The camera-based optical flow system acts as an intermediary that processes visual information to extract reliable yaw rate measurements. Instead of directly fusing raw inertial and GNSS data, the patent uses optical flow as an intermediate representation that is less susceptible to thermal drift and signal reception problems, thereby maintaining precision while utilizing multiple information sources.
Solution Approach 2:
The patent substitutes the mechanical inertial measurement system with an optical-based measurement system for yaw rate estimation. By replacing reliance on physical sensors prone to thermal drift with optical flow computation from camera images, the system maintains measurement precision while still utilizing multiple information sources through sensor fusion.
3Speed
If inertial sensors are used to measure yaw rate directly, then the response speed is fast, but the measurement precision deteriorates due to significant uncertainties from thermal drift
Solution Approach 1:
The system uses camera-based optical flow measurements as feedback to correct the fast but drift-prone inertial sensor readings. The optical flow provides periodic verification and correction of the rapid inertial measurements, maintaining both the speed response of inertial sensors and the measurement precision required for accurate yaw rate estimation.
Solution Approach 2:
The patent merges the fast response capability of inertial sensors with the measurement precision of optical flow-based methods. By combining these two complementary approaches through sensor fusion, the system achieves both rapid yaw rate detection and sustained measurement accuracy, overcoming the limitations of using either system alone.
Data Source
AI summary
A method for ascertaining a highly accurate piece of yaw rate information for controlling a vehicle is provided. The method includes ascertaining a first yaw rate estimated value of the vehicle based on a fusion of sensor data of an inertial sensor, a GNSS sensor, a wheel velocity sensor and/or a steering angle sensor; ascertaining a second yaw rate estimated value of the vehicle by an evaluation of sensor data of a camera assigned to the vehicle, which optically detects the surroundings of the vehicle; carrying out a correction of the first yaw rate estimated value with the aid of the second yaw rate estimated value to ascertain a corrected yaw rate estimated value; and outputting the corrected yaw rate estimated value as a piece of yaw rate information to generate a control signal for controlling the vehicle.