Hybrid Vehicle Trajectory Control for Unmodeled Dynamics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle control systems face challenges in accurately tracking vehicle trajectories due to sensitivity to unmodeled dynamics, parameter variations, and non-linearities, particularly when using model-based controllers alone.
Innovation Solution
A vehicle control system that combines a model-based controller with a neural network controller, where the neural network generates additional control signals based on pre-trained tuning parameters and predicted states, and these signals are combined with model-based signals to operate vehicle actuators, allowing for online tuning and improved non-linear control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a model-based controller is used for vehicle control, then the control system has a clear theoretical framework and structure, but the system exhibits sensitivity to unmodeled dynamics and parameter variations
Solution Approach 1:
The patent combines a model-based controller with a neural network controller into a hybrid control system. The model-based controller provides the theoretical framework and structure, while the neural network controller compensates for unmodeled dynamics and parameter variations. The controllers work in parallel with their outputs combined, resolving the contradiction between structural clarity and tracking accuracy.
Solution Approach 2:
The neural network acts as an intermediary that bridges the gap between the model-based controller's theoretical framework and the actual vehicle dynamics. It learns the discrepancies between the model and real system behavior, providing corrective control signals that improve trajectory tracking without disrupting the overall control structure.
2Reliability
If a neural network controller is used alone, then the system can handle non-linearities and unmodeled dynamics, but the system requires extensive training data and computational resources
Solution Approach 1:
By merging the neural network controller with a model-based controller, the system leverages the strengths of both approaches. The model-based controller handles the majority of control tasks using its theoretical framework, while the neural network focuses only on compensating for non-linearities and unmodeled dynamics, reducing training requirements and computational burden.
Solution Approach 2:
Instead of using a complete neural network controller that would require extensive training, the patent applies partial action by using the neural network only for specific compensation tasks. The neural network processes only the error signals and specific state variables needed for compensation, reducing computational resources and training data requirements while maintaining improved control performance.
3Productivity
If traditional control methods are used, then the system is computationally efficient and easy to implement, but the system lacks accuracy in trajectory tracking under varying conditions
Solution Approach 1:
The patent merges traditional model-based control with neural network control to achieve both computational efficiency and high accuracy. The model-based controller provides computationally efficient baseline control, while the neural network adds accuracy for trajectory tracking under varying conditions. The combined approach maintains real-time performance while improving precision.
Data Source
AI summary
A vehicle control system for automated driver-assistance includes a model-based controller that generates a first control signal to alter an operation of a plurality of actuators of a vehicle based on a reference trajectory for the vehicle, and a present state of the vehicle and actuators. The vehicle control system further includes a neural network controller that generates a second control signal to alter the operation of the actuators of the vehicle based on a reference trajectory for the vehicle, and the present state of the vehicle and actuators. The vehicle control system further includes a combination module that combines the first control signal and the second control signal to operate the actuators based on a combined signal.