Reduced-DOF Robot Control With Partial Actuator Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Robotic devices with multiple degrees of freedom are challenging to control simultaneously due to a mismatch between user input capabilities and the robot's complex movements, leading to difficulties in training and operation, especially when controlling multiple spatial dimensions.
Innovation Solution
An adaptive controller using a neuron network is implemented, allowing for supervised learning by training separate actuator subsets during trials, with the controller generating control signals for multiple degrees of freedom based on user input and learning parameters, enabling the robot to perform target actions without continuous user input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple degrees of freedom are controlled simultaneously, then the robot can perform complex movements in multiple spatial dimensions, but the control complexity increases significantly making it difficult for users to operate
Solution Approach 1:
The control of multiple degrees of freedom is segmented into separate training iterations, where each iteration focuses on training a specific subset of actuators (e.g., one degree of freedom at a time). This breaks down the complex simultaneous control task into manageable segments that users can master sequentially, reducing cognitive load while maintaining the robot's full movement capability.
Solution Approach 2:
The system performs preliminary training actions for each degree of freedom before combining them for full operation. During training iterations, the robot learns to control individual actuators independently first, then integrates these pre-trained control patterns to achieve coordinated multi-degree-of-freedom movements, making the overall system easier to operate.
2Reliability
If all actuators are trained simultaneously, then the robot can learn complex coordinated movements, but the training time and computational resources required increase significantly
Solution Approach 1:
The training process is segmented into multiple iterations, with each iteration dedicated to training a specific subset of actuators rather than all actuators simultaneously. This segmentation reduces the computational complexity and training time for each iteration while maintaining the ability to learn coordinated movements through sequential integration of trained actuator subsets.
Solution Approach 2:
The system trains only the necessary subset of actuators in each iteration rather than all actuators continuously. This partial action approach reduces training time and computational resources per iteration, while the cumulative effect of multiple iterations achieves comprehensive coordinated movement capability.
3Ease of operation
If a simple control interface is used, then the user experience is improved, but the robot's ability to perform complex multi-degree-of-freedom tasks is limited
Solution Approach 1:
The system introduces an intermediary adaptive controller that translates simple user inputs into coordinated control signals for multiple actuators. This intermediary layer handles the complexity of multi-degree-of-freedom coordination internally, allowing users to interact with a simple interface while the robot performs complex tasks through the adaptive controller's learned actuator coordination patterns.
Solution Approach 2:
The adaptive controller performs self-learning and self-adjustment during training iterations, automatically optimizing the coordination between actuators without requiring complex user intervention. This self-service capability enables the system to develop complex task performance abilities while maintaining interface simplicity, as the controller autonomously manages the complexity of multi-actuator coordination.
Data Source
AI summary
Apparatus and methods for training and controlling of, for instance, robotic devices. In one implementation, a robot may be trained by a user using supervised learning. The user may be unable to control all degrees of freedom of the robot simultaneously. The user may interface to the robot via a control apparatus configured to select and operate a subset of the robot's complement of actuators. The robot may comprise an adaptive controller comprising a neuron network. The adaptive controller may be configured to generate actuator control commands based on the user input and output of the learning process. Training of the adaptive controller may comprise partial set training. The user may train the adaptive controller to operate first actuator subset. Subsequent to learning to operate the first subset, the adaptive controller may be trained to operate another subset of degrees of freedom based on user input via the control apparatus.