Vehicle Control Data Learning for Preference-Based Drive Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems require extensive manual effort from skilled workers to set up and adapt operation amounts of electronic devices based on vehicle states, leading to inefficiencies and increased man-hours.
Innovation Solution
A vehicle control data generation method using reinforcement learning to adjust the relationship between vehicle states and action variables, providing rewards based on user preferences for elements like acceleration response, noise, and energy efficiency, thereby automating the optimization process and reducing manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual adaptation of filter and operation amounts is performed by skilled workers, then appropriate operation amounts are set, but a great number of man-hours are required
Solution Approach 1:
The system performs self-adaptation through reinforcement learning, where the vehicle controller automatically learns optimal operation amounts for electronic devices based on vehicle states and user preferences, eliminating the need for manual adjustment by skilled workers
Solution Approach 2:
The patent replaces the mechanical/manual adjustment process with an automated computational system using reinforcement learning algorithms that process sensor data, preference variables, and reward signals to determine optimal control parameters
2Extent of automation
If reinforcement learning is used to automatically set operation amounts, then manual intervention is reduced, but the system complexity increases
Solution Approach 1:
The vehicle controller performs multiple functions including normal vehicle control, sensor data processing, preference variable interpretation, reward calculation, and reinforcement learning-based adaptation, consolidating these diverse functions into a single integrated system
Solution Approach 2:
The system dynamically adjusts operation amounts by changing control parameters based on learned relationships between vehicle states, user preferences, and reward signals, allowing flexible adaptation without hardware modifications
Data Source
AI summary
A vehicle control data generation method is provided. A preference variable indicates a relative preference of a user for two or more requested elements that include at least two of three requested elements including a requested element indicating a high acceleration response of a vehicle, a requested element indicating at least one of vibration and noise of the vehicle is small, and a requested element indicating a high energy use efficiency. The reward calculating process includes a changing process that changes a reward provided when a characteristic of the vehicle is a predetermined characteristic in a case where a value of the preference variable is a second value such that the changed reward differs from the reward provided when the characteristic is the predetermined characteristic in a case where the value of the preference variable is a first value.