Brain-Like Robot Motion Control for Trial-and-Error Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robots struggle to effectively combine environmental and memory information for decision-making, cannot flexibly select and execute operations, and lack autonomous trial-and-error learning capabilities, especially in complex tasks with high degrees of freedom.
Innovation Solution
A brain-like decision-making and motion control system incorporating an active and automatic decision-making module, evaluation, memory, perceptual, and compound control modules, which process multi-modal information, adjust operations based on feedback, and enable trial-and-error learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If inverse kinematics is used to solve trajectory for high-degree-of-freedom systems, then motion planning can be achieved, but computational complexity becomes very high and solutions are not unique
Solution Approach 1:
The patent segments the high-degree-of-freedom system into multiple subsystems with fewer degrees of freedom each. By dividing the complex motion planning problem into smaller sub-problems that can be solved independently, the computational complexity is reduced while maintaining the overall motion planning capability of the complete system.
Solution Approach 2:
The patent transforms the complex high-dimensional motion planning problem by introducing a hierarchical control structure. The control architecture operates at multiple levels, where higher levels handle strategic decision-making and lower levels handle execution, effectively reducing the dimensionality of the computational problem at each level.
2Adaptability or versatility
If the robot independently selects targets and operations with integration of internal state, external environment, and historical information, then decision-making capability is improved, but the system complexity increases
Solution Approach 1:
The patent segments the decision-making system into modular functional components including perceptual modules for sensing, memory modules for storing historical information, decision modules for selecting targets and operations, and execution modules for carrying out actions. This modular architecture enables comprehensive decision-making while managing system complexity through organized functional separation.
Solution Approach 2:
The patent implements universal decision-making modules that can handle multiple types of decisions across different contexts. These modules integrate internal state, external environment, and historical information through standardized interfaces, allowing the same core architecture to serve multiple functions and reduce overall system complexity.
3Adaptability or versatility
If the robot needs to flexibly select, switch and control each sub-action/meta-action from different initial states, then adaptability to changes and interference is improved, but control complexity increases
Solution Approach 1:
The patent implements a dynamic control architecture where the control structure can adapt its organization and behavior based on the current task requirements and environmental conditions. The system can dynamically switch between different control modes and reconfigure control parameters, enabling flexible operation from different initial states without requiring a completely rigid control structure.
Solution Approach 2:
The patent introduces a hierarchical dimension to the control architecture, organizing control into multiple levels from meta-action planning to sub-action execution. This hierarchical structure allows the system to manage complexity by handling different aspects of control at appropriate levels, enabling flexible switching and coordination of actions without overwhelming control complexity at any single level.
4Adaptability or versatility
If the robot performs autonomous trial-and-error learning or imitation learning to learn new motions and trajectories, then learning capability is improved, but computational and data processing requirements increase
Solution Approach 1:
The patent implements preliminary structuring of learning through pre-defined action spaces, predefined task templates, and pre-organized training data structures. By preparing the learning framework in advance with appropriate constraints and guidelines, the system can perform trial-and-error and imitation learning more efficiently, reducing the computational burden during actual learning execution.
Solution Approach 2:
The patent introduces intermediary components such as memory modules that store learned experiences and knowledge bases that provide guidance during learning. These intermediaries act as buffers and organizers between the learning processes and the execution systems, managing the complexity of computational and data processing requirements by structuring and caching information in accessible formats.
Data Source
AI summary
A brain-like decision-making and motion control system is disclosed, this system comprises an active decision-making module, an automatic decision-making module, an evaluation module, a memory module, a perceptual module, a compound control module, an input channel module, an output channel module, and a controlled object module. The three working modes supported by the system include an active supervision mode, an automatic mode, and a feedback-driven mode, which enables the robot to make autonomous decisions to select targets and operations and delicately control actions in the process of interacting with the environment, and can make the robot learn new operations by trial and error, imitation, demonstration, with the flexibility to adapt to the complex task and the environment.