Motor Controller Inversion for Robot Torque Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot control schemes that use independent-joint controllers with constant feedback gains fail to perform well under varying speeds and payloads due to their inability to update gains, leading to perturbations and poor performance in non-linear systems with changing inertial and gravity effects.
Innovation Solution
The implementation of a motor-controller inversion (MCI) block that inverts the dynamics of a robot's built-in motor and controller, using a combination of open-loop and closed-loop systems with feedback to achieve robustness, allowing for the cancellation of motor and controller dynamics and enabling torque control even in robots that only accept position or velocity commands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If independent-joint controllers with constant feedback gains are used, then the control scheme is simple and easy to implement, but the performance deteriorates under varying speeds and payloads due to inability to update gains
Solution Approach 1:
The patent transforms the static, constant feedback gains into dynamic, time-varying gains that adapt to changing operating conditions. The feedback gains are updated in real-time based on the current state of the robot system, allowing the controller to maintain optimal performance across varying speeds and payloads while preserving the fundamental simplicity of the independent-joint control architecture.
2Adaptability or versatility
If torque commands are applied to robots, then advanced control algorithms can be tested and validated, but robots with built-in controllers only accept position or velocity commands
Solution Approach 1:
The patent introduces an intermediary control layer that sits between the user's torque commands and the robot's built-in position/velocity controller. This intermediary layer translates torque commands into equivalent position or velocity commands that the built-in controller can execute, while still allowing researchers to test and validate advanced torque-based control algorithms on robots with limited command interfaces.
3Device complexity
If joint controllers act independently, then each joint can be controlled separately with simple controllers, but perturbations are produced at neighboring joints due to lack of coordination
Solution Approach 1:
The patent implements a feedback mechanism where each joint controller not only controls its own joint but also receives information about the state and control actions of neighboring joints. This allows the independent joint controllers to compensate for perturbations caused by adjacent joint movements, maintaining simple controller structures while achieving coordinated system-level performance through information sharing and adaptive gain adjustment.
Data Source
AI summary
Systems and methods are presented that cancel the dynamics of a motor and a joint controller in the presence of communication time delays, measurement noise, and controller parameter uncertainties inherent in a robot system. A cancellation system includes feedback of the measured output Ê of a controller block C. A first summing node (Σ1) subtracts Ê from the desired voltage Ed to determine a voltage error, which is fed into a G(s) block. A second summing node (Σ2) adds the output of the G(s) block to the desired voltage Ed to generate the signal Ē, which is fed into a non-robust controller inversion block C−1. The block C−1 outputs a value μ, which is fed into the block C. The block C outputs the actual voltage E, which is input into a motor block M. Under perfect conditions, the voltage error is zero, in which case the input to block G(s) is zero.