Robotic Manipulator Teleoperation via Joint Velocity Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic manipulator control systems require highly customized and expensive hardware, such as haptic gloves, to operate dexterous robots with many degrees-of-freedom, limiting usability and accessibility.
Innovation Solution
A method for controlling robot manipulators using a teleoperation device with fewer degrees-of-freedom by formulating an optimization problem to determine joint velocities, allowing control of robots with a lower DoF device through a least squares solver and machine learning techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If highly customized hardware such as haptic gloves is used to control dexterous robots, then the robot can be operated with full degrees-of-freedom, but the system becomes expensive and less accessible
Solution Approach 1:
The patent creates a virtual model (digital twin) of the robotic hand with full degrees of freedom, which is controlled through a simplified teleoperation device. The virtual model replicates the complex robot's behavior and responses, allowing users to interact with the complex system through a simpler interface without requiring physically complex hardware like haptic gloves.
Solution Approach 2:
The patent introduces an intermediary control system that translates inputs from a simple teleoperation device into coordinated movements of multiple robotic joints. This intermediary layer (including the optimization controller and reinforcement learning model) acts as a mediator between the simple user input and the complex robot actuation, eliminating the need for complex input hardware.
2Device complexity
If a teleoperation device with fewer degrees-of-freedom is used, then hardware cost and complexity are reduced, but controlling a robot with many degrees-of-freedom becomes more difficult
Solution Approach 1:
The patent transforms the control problem by changing parameters through optimization. The reinforcement learning model learns optimal mappings from limited teleoperation device inputs to multiple robot joint commands. The system dynamically adjusts control parameters and weightings to achieve accurate robot control despite the input device having fewer degrees of freedom than the robot.
Solution Approach 2:
The patent performs preliminary training of the reinforcement learning model offline before actual operation. During this preliminary phase, the system learns the optimal control strategies and mappings from simple inputs to complex robot movements. This pre-computed knowledge is then applied during operation, making the control process straightforward and easy to use without requiring the operator to manually coordinate multiple joints.
3Measurement precision
If optimization problems are solved in real-time to determine joint velocities, then control accuracy is improved, but computational resources and processing time increase
Solution Approach 1:
The patent performs the computationally intensive optimization and learning processes in advance. The reinforcement learning model is trained offline to capture the optimal control policies. During real-time operation, the system only needs to query the pre-trained model and execute the determined joint velocities, avoiding real-time optimization computation while maintaining high control accuracy.
Solution Approach 2:
The patent replaces real-time computational optimization with a pre-trained machine learning model that provides direct control mappings. Instead of solving optimization problems in real-time, the system uses the learned policy from the reinforcement learning model to directly determine joint velocities from teleoperation inputs, significantly reducing computational processing time while maintaining accuracy.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling a robot manipulator that has a plurality of joints. One of the methods includes obtaining a control input that comprises one or more velocity values that specify a target velocity of a reference point in a given coordinate frame; determining a respective joint velocity for each of the plurality of joints by generating a solution to an optimization problem formulated from the control input; and controlling the robot manipulator, including causing the plurality of joints of the robot manipulator to move in accordance with the respective joint velocities to approximate the control input.