Autonomous Driving Personalization Function Learning and Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems face challenges in ensuring safety and comfort due to conflicts between actions requiring different processing times, leading to potential wrong selections and discomfort for drivers, as they may feel the vehicle brakes earlier or later than desired.
Innovation Solution
A driving control apparatus and method that incorporates a personalization function storage section and updating section, allowing for the authentication and learning of driver-specific preferences, which are used to control vehicle actions based on detection results, thereby optimizing braking and other driving actions to match driver habits and customs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous driving control determines actions based on multiple detection results requiring different processing times, then comprehensive decision-making is achieved, but conflicts between actions may occur leading to wrong selections
Solution Approach 1:
The system performs preliminary classification of actions into reflex actions (requiring immediate response) and deliberate actions (allowing processing time). By pre-defining these categories and their handling priorities, the system resolves conflicts before they occur, ensuring safety-critical reflex actions are executed promptly while deliberate actions are processed comprehensively.
Solution Approach 2:
The action control is segmented into separate processing paths: one for reflex actions and another for deliberate actions. This segmentation allows the system to handle different types of decisions through appropriate mechanisms, preventing conflicts between actions with different timing requirements and improving overall reliability.
2Ease of operation
If autonomous driving control uses standardized actions for all drivers, then system simplicity is maintained, but driver comfort and preference tailoring cannot be achieved
Solution Approach 1:
The system changes parameters by introducing personalization functions that adjust action characteristics based on individual driver preferences. These functions modify parameters such as braking intensity, acceleration rates, and route selection criteria to match each driver's comfort preferences, enhancing ease of operation without fundamentally changing the control architecture.
Solution Approach 2:
The system creates copies of the basic autonomous driving control functionality, tailored to each driver's preferences through personalization functions. Each driver receives a customized version of the control system that maintains the core safety and decision-making logic while adjusting behavioral parameters to match individual comfort levels.
3Adaptability or versatility
If personalization functions are learned and stored for multiple drivers, then driver-specific comfort is achieved, but data management and verification complexity increases
Solution Approach 1:
The system implements self-service through automated learning of personalization functions from driver behavior data. The learning section automatically acquires and stores personalization functions for multiple drivers without manual intervention, and the verification section automatically validates these functions before use, reducing data management complexity while maintaining high adaptability.
Solution Approach 2:
The system uses feedback mechanisms where the verification section validates learned personalization functions before they are applied in autonomous driving control. This feedback loop ensures data quality and system reliability, managing the complexity of storing and using multiple driver-specific functions through automated verification processes.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to a driving control apparatus, a driving control method, and a program that can safety update a personalization function that has been learned to match habits and customs of each driver on the basis of a driving operation of a driver and detection results of various sensors. A personalization function for each driver is found by learning on the basis of driving operation conducted by a driver through manual driving and detection results of various sensors provided on a vehicle body. Verification simulation is carried out by using the found personalization function, and when safety thereof is confirmed, the personalization function is updated as a new personalization function. Autonomous driving reflecting habits and customs of the driver can be realized by correcting an action command found during autonomous driving on the basis of the detection results with the personalization function, thereby realizing safe and comfortable autonomous driving. The present disclosure is applicable to motor vehicles that drive autonomously.