Efficient adaptive control improves robot arm manipulation in real time
RoKON: Efficient and Adaptive Data-Enabled Predictive Control for Robotic Manipulation
Robotics
Summary
Robots often need to adjust their movements based on changing conditions, like friction on a surface. This paper introduces a new method that helps a robot arm predict and adapt its actions quickly without needing a detailed model of the robot’s mechanics. The authors combine smart mathematical techniques with a way to keep important recent and older data. They tested their method by having a robot push a block across surfaces with different friction, showing it can track and control motions effectively in real time.
What this means in practice
- •For robotics engineers: Control robot arms to adapt in real time to varying object or surface properties without relying on pre-built models.
- •For industrial automation teams: Improve robotic manipulators in factories to handle changing conditions, like tool wear or material inconsistencies, during pick-and-place or assembly tasks.
Authors
Akshun Sharma, Zhongqi Wei, Buqing Ou, Frederike Dümbgen
Abstract
Data-enabled predictive control has recently emerged as a promising control paradigm, enabling a transparent optimal-control formulation without the need to learn an explicit dynamics model. This paper presents a number of extensions paving the path towards its use in real-time robotics tasks. Following recent work, we utilize a nonlinear predictor based on Gaussian kernels to represent the relationship between input and output data. We propose a damped Gauss-Newton method that enables real-time control by efficiently exploiting the least-squares structure of the control problem. In addition, we introduce a hybrid buffer management strategy that incorporates informative online data while retaining relevant offline data, allowing the predictor to account for changes in the dynamics. We demonstrate the tracking performance and computational suitability of the proposed method in MuJoCo using a Franka Research 3 arm pushing a block across surface, without prior knowledge of its changing friction values.