Human Motion Descriptor Robot Control System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for controlling robot motion using human demonstration are limited by the complexity of sensing and instrumentation, making them time-consuming and unsuitable for dynamic environments, and rely on special instrumentation and off-line optimization, which is not suitable for real-time or reactive control.
Innovation Solution
A unified, task-oriented dynamic control system that uses human motion descriptors to generate robot joint commands, enforcing kinematic and dynamic constraints through a hierarchical framework, allowing for real-time control and adaptation in both static and dynamic environments without requiring special instrumentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If marker based optical systems or magnetic systems are used to record human motion, then motion data can be captured, but the sensing and instrumentation complexity increases and the process becomes time-consuming
Solution Approach 1:
The patent extracts only the essential motion information (task descriptors) from human movement and transfers it directly to the robot, eliminating the need for complex marker-based optical systems or magnetic systems. The motion capture process is simplified to recording only relevant task-level parameters rather than full kinematic data.
Solution Approach 2:
Instead of physically replicating complex motion capture instrumentation, the patent creates a computational model that copies human motion patterns through task descriptors. The robot learns from demonstrated tasks by processing simplified motion representations rather than requiring identical sensing equipment.
2Reliability
If off-line optimization procedures are used to generate robot motion, then kinematic and dynamic constraints can be satisfied, but the control process is not suitable for real-time or reactive control in dynamic environments
Solution Approach 1:
The patent performs preliminary computation by pre-calculating task descriptors from human demonstrations and storing them for later use. During real-time operation, the robot retrieves and executes pre-processed motion commands rather than performing optimization calculations, enabling fast response while maintaining constraint satisfaction.
Solution Approach 2:
The system transitions from static off-line optimization to dynamic real-time control by implementing a hierarchical framework where task-level descriptors guide high-level motion planning while lower-level controllers handle real-time adjustments. This allows the robot to adapt to dynamic environments while maintaining reliability.
3Measurement precision
If special instrumentation is used for human motion capture, then accurate motion data can be obtained, but the system is not suitable for use outside the laboratory
Solution Approach 1:
The patent develops a universal task descriptor framework that can capture essential motion information using simple, inexpensive sensors that work in various environments. The system is designed to function both in controlled laboratory settings and in unstructured real-world environments, making it versatile and broadly applicable.
Solution Approach 2:
The system uses the robot's own sensors and actuators to capture and execute motion, eliminating the need for external specialized instrumentation. The robot performs self-calibration and adapts to different environments without requiring dedicated motion capture equipment, enabling deployment outside the laboratory.
Data Source
AI summary
A control system and method generate torque comments for motion of a target system in response to observations of a source system. Constraints and balance control may be provided for more accurate representation of the motion as replicated by the target system.