Smart Gains Model for Autonomous Vehicle Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic driving technologies face challenges such as high computational complexity, the 'black-box' problem in neural networks, and the inability to effectively manage multi-objective control for safe driving, comfortable riding, and energy efficiency, leading to difficulties in man-machine sensory fusion and ethical decision-making in emergency situations.
Innovation Solution
The 'Smart Gains' model employs a machine learning approach using a maximum probability Gaussian process to learn driving skills from human experience, enabling optimized control of automatic driving vehicles through a decision information module, data module, and driving module that integrates machine consciousness and prior knowledge for closed-loop control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional neural network algorithms are used for automatic driving control, then the system can learn from data, but the computational complexity becomes huge and hardware overhead increases significantly
Solution Approach 1:
The patent extracts only the essential learning capabilities needed for automatic driving control from complex neural networks, implementing a simplified control algorithm that achieves the necessary adaptability without the computational burden of full neural network processing
Solution Approach 2:
The patent replaces the mechanical neural network computation system with an optimized control algorithm that achieves similar learning outcomes through more efficient mathematical operations, reducing hardware requirements and computational complexity
2Adaptability or versatility
If traditional neural network models are used, then the system can process driving data, but the threshold values are defined artificially and the black-box problem cannot be solved
Solution Approach 1:
The patent introduces an intermediary layer between data input and control output that provides interpretable intermediate representations, allowing the system to maintain data processing capability while making the decision-making process transparent and understandable
Solution Approach 2:
The patent changes the parameter representation from artificial threshold values to continuously adjustable control parameters that can be optimized through learning, enabling both data-driven adaptation and interpretability through meaningful parameter semantics
3Adaptability or versatility
If deep learning with more hidden layers is used to improve learning capability, then the model becomes more powerful, but the calculation becomes more complex and the black-box problem worsens
Solution Approach 1:
The patent applies partial learning - using only the essential learning capabilities needed for automatic driving control rather than attempting to implement full deep learning, achieving sufficient adaptability with reduced computational complexity
4Reliability
If rule-based fuzzy inference is used for automatic driving control, then the system can handle uncertainty, but it cannot effectively manage multi-objective control for safety, comfort, and energy efficiency simultaneously
Solution Approach 1:
The patent merges the uncertainty handling capabilities of fuzzy logic with the multi-objective optimization capabilities of modern control theory, creating a unified control framework that simultaneously addresses safety, comfort, and energy efficiency while maintaining interpretability
Data Source
AI summary
A control method of automatic driving imported “Smart Gains” model can bypass the problem caused by high levels of complexity that currently trouble automatic driving control systems. The knowledge generated by a Gaussian process machine learning model with the maximum probability can carry out the closed-loop control of automatic driving with a given trajectory, can solve the nonlinear adjustment problem of the actuators of the automatic driving vehicle, as well as the optimization control problem of the randomness of the control object. This feature can also make the automatic driving vehicle run smoothly, save energy, be comfortable and fast, and achieve automatic driving above Level 4.