Autonomous Driving Control Gain Tuning for Map Position Deviations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face challenges in accurately estimating their position, particularly in changing lane sections, leading to unexpected vehicle behavior due to differences between map data and actual road conditions, which affects steering control and tracking performance.
Innovation Solution
A control device equipped with behavior detection sensors and external detection sensors that estimate the vehicle's current position on map data, calculate correction amounts, and adjust control gain based on detected deviations to improve tracking performance and prevent unexpected behavior.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If weighting data is changed in accordance with each component (lateral direction and yaw direction) to estimate vehicle position, then position estimation can be performed for each direction, but deviation detection becomes inconsistent between directions leading to unexpected vehicle behavior
Solution Approach 1:
The patent applies local quality by using different weighting strategies for different components (lateral direction vs. yaw direction) when estimating vehicle position. For lateral direction, camera-recognized position information is weighted, while for yaw direction, sensor-detected vehicle behavior information is weighted. This localized adaptation to different measurement characteristics resolves the contradiction by optimizing position estimation accuracy for each direction while maintaining overall system reliability through consistent control execution.
Solution Approach 2:
The patent implements dynamics by dynamically adjusting the weighting of position information and vehicle behavior information based on the specific direction being estimated. The control execution unit changes control characteristics dynamically according to the estimated position and detected vehicle behavior, allowing the system to adapt to varying conditions in different directions while maintaining consistent and predictable vehicle response.
2Measurement precision
If camera-recognized position information is weighted for lateral direction and sensor-detected vehicle behavior information is weighted for yaw direction, then direction-specific position estimation is achieved, but map data deviations cause inconsistent deviation detection between directions
Solution Approach 1:
The patent applies feedback by continuously monitoring both the estimated position and actual vehicle behavior, then using this information to adjust control execution. The control execution unit receives feedback from position estimation and vehicle behavior detection, and dynamically adjusts control characteristics to compensate for map data deviations. This feedback mechanism ensures consistent deviation detection across different directions by comparing estimated position with actual vehicle response.
Solution Approach 2:
The patent implements parameter changes by modifying the weighting parameters for position information and vehicle behavior information based on the estimation direction and detected deviations. When map data deviations are detected, the system changes the weighting parameters to rely more on actual sensor measurements rather than map data, thereby maintaining accurate deviation detection across all directions despite using different primary information sources for different components.
3Productivity
If control gain is adjusted based on correction amount to improve tracking performance, then vehicle returns to target path faster, but control input changes may cause unexpected behavior if not properly coordinated
Solution Approach 1:
The patent applies preliminary action by calculating and preparing correction amounts for position deviations before executing control adjustments. The control execution unit proactively adjusts control characteristics based on predicted needs rather than reacting to large deviations. This preliminary adjustment ensures that control gain changes are smooth and coordinated, improving path tracking speed while maintaining predictable vehicle behavior by avoiding sudden, uncoordinated control input changes.
Solution Approach 2:
The patent uses feedback to continuously monitor vehicle position and behavior during control execution. The control execution unit receives real-time feedback on the effectiveness of control adjustments and dynamically modifies control characteristics to maintain smooth, coordinated responses. This feedback mechanism ensures that increased control gain for faster tracking does not cause unexpected behavior, as the system continuously adapts to maintain predictable vehicle response.
Data Source
AI summary
A control device for causing a vehicle to travel by autonomous driving based on a target path calculated based on a current position on map data of a vehicle estimated by a map data, a behavior detection sensor, and an external detection sensor, wherein a current position of the vehicle is estimated by correcting a position on map data of the vehicle in accordance with a relative position of the vehicle on a roadway, a correction amount at that time is obtained, a change in a correction amount is predicted by map data and an external detection sensor, and a control gain in a control input when the vehicle is caused to travel by autonomous driving is changed in accordance with the change, thereby changing a characteristic of the traveling control.