Task-Space Disturbance Observer for Stable Robot Admittance Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Admittance control in robots, especially those with position control, faces stability issues due to internal loop bandwidth, time delays, and model errors when interacting with external environments.
Innovation Solution
A task space outer-loop integrated disturbance observer is implemented outside the position and velocity control loop, which estimates disturbances by integrating velocity command values, measured velocities, an inverse model of the velocity control system, and force/torque sensor measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If admittance control is implemented on position control hardware, then human-robot interactive task performance is improved, but contact stability deteriorates due to deviations from ideal reference model
Solution Approach 1:
The patent implements a disturbance observer that continuously monitors the difference between actual and nominal system behavior, generating feedback signals to compensate for model deviations. The observer uses measured velocity and force/torque data to estimate disturbances, which are then fed back to correct the control output, thereby maintaining contact stability while preserving interactive task performance.
Solution Approach 2:
The disturbance observer acts as an intermediary component between the admittance controller and the position control loop. It processes the discrepancy between actual and nominal dynamics and generates compensating signals that mediate the interaction, allowing the system to maintain stability without compromising the admittance control performance.
2Productivity
If admittance control is implemented on position control hardware, then human-robot interactive task performance is improved, but admittance rendering accuracy deteriorates due to internal loop bandwidth limitations
Solution Approach 1:
The disturbance observer provides continuous feedback compensation that corrects for bandwidth limitations in the internal control loop. By monitoring the deviation from nominal behavior and generating real-time compensation signals, the system maintains higher admittance rendering accuracy than would be achievable with the limited bandwidth hardware alone.
Solution Approach 2:
The patent effectively changes the dynamic parameters of the control system by adding the disturbance observer, which modifies the frequency response characteristics. The observer compensates for the bandwidth limitations of the position control hardware, allowing the system to achieve admittance rendering accuracy that would otherwise require higher bandwidth hardware.
3Productivity
If admittance control is implemented on position control hardware, then human-robot interactive task performance is improved, but model accuracy deteriorates due to time delay and internal loop bandwidth
Solution Approach 1:
The disturbance observer continuously monitors the discrepancy between actual and nominal system responses, providing real-time feedback that compensates for time delays and bandwidth limitations. This feedback mechanism maintains model accuracy by dynamically adjusting for deviations caused by hardware constraints.
Solution Approach 2:
The disturbance observer performs preliminary compensation for expected model inaccuracies by estimating disturbances based on measured deviations. This preliminary action corrects for time delays and bandwidth effects before they significantly degrade model accuracy, maintaining reliable interaction performance.
Data Source
AI summary
The present disclosure relates to a task space outer-loop integrated disturbance observer, and a robot including the same. The task space outer-loop integrated disturbance observer is implemented on an outside of a position and velocity control loop in a task space, and the task space outer-loop integrated disturbance observer acquires a disturbance estimate value by integrating a velocity command value, measured velocity value, an inverse model of velocity control system, and a measured force/torque sensor (F/T sensor) value. The disturbance estimate value is expressed in Equation 1: {circumflex over (D)}v(s)=Q(s)[n−1(s)m(s)−i(s)]+A(s)[1−Q(s)]m(s) wherein {circumflex over (D)}v(s) is the disturbance estimate value, Q(s) is a “Q” filter, Dn(s) is a nominal model, Vm(s) is a measured velocity value, and Vi(s) is a velocity command value, A(s) is an admittance target value, and Fm(s) is a measured force value.