Risk-sensitive interaction control in uncertain manipulation tasks

Abstract

Abstract-Manipulation tasks are a great challenge for robots due to the uncertainty arising from unstructured environments. In this paper we propose a novel control scheme for contact tasks based on risk-sensitive optimal feedback control. It provides a systematic approach to adjust the trade-off between motion and force control under uncertainty. Following a previously acquired task model, the proposed approach provides both a variable stiffness solution and a motion reference adaptation. This control scheme achieves increased adaptability under previously unseen environmental variability. An implementation on a robotic manipulator validates the applicability and adaptability of the proposed control approach in two different manipulation tasks.

Gaussian processes for dynamic movement primitives with application in knowledge-based cooperation

“…Here we focus on the application to multi agent systems, e.g. variance-based impedance control [17].…”

Distributed adaptive impedance control of networked Lagrangian systems with neighborhood interaction feedback

“…21 A variable impedance controller based on the reinforcement learning (RL) algorithm PI 2 is used in high degree of freedom robotic manipulator 22 and biological motor. 23 Furthermore, a risk-sensitive optimal feedback control scheme 24 was proposed in manipulator to adjust the trade-off between motion and force control under uncertainty. Then Howard et al 25 investigated the comparison of several robotic control approaches and imitated the behavior of humans by variable impedance actuators.…”

Passivity Filter for Variable Impedance Control

“…As a result, variable impedance control (VIC) has recently attracted more and more attention. VIC has been explored for challenging objectives, for example to deal with explosive movements [6], to optimize the performance of hammering tasks [7], allow risk-sensitive interactions [8] or maximize robot links velocities [9]. VIC has been implemented using reinforcement learning [10], as well as adaptive approaches for human-robot collaboration based on the estimation of the human arm stiffness, from the derivatives of force and position [11], or from the measurement of muscle activity with electromyography [12].…”