Optimize and coordinate multiple DMPs under constraints to achieve a collaborative manipulation task - Ali H. Kordia and Francisco S. Melo

Oct 3, 2026

This paper tackles how robots or multiple robots, equipped with a library of pre-learned dynamic movement primitives (DMPs), can generate multiple simultaneous coordinated motions to complete one collaborative manipulation task. It extends a single-manipulator DMP approach to coordinated execution across multiple manipulators or multiple robots, building the joint trajectory from initially learned DMPs and optimizing it under task-specific constraints. The method is demonstrated in simulation and with a Baxter robot in both simulated and real settings.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Abstract

Introduction

Optimize and coordinate multiple DMPs under constraints to achieve a collaborative manipulation task Ali H. Kordia 1 Francisco S. Melo2 Abstract—This paper addresses a significant challenge in achieving collaborative tasks; how can a robot or multiple robots, endowed with a library of pre-learned primitive movements, generate multiple simultaneous coordinated robotic movements, adapting and optimizing those in the library, to complete one collaborative task? This work can thus be seen asa follow-up to the work with a motion presented as a dynamic movement primitive (DMP) that now considers collaborative tasks and the existence of multiple robots/manipulators. Specifically, we start with a simple task using one DMP and extend it to accommodate the coordinated execution of multiple DMPs in robots with multiple manipulators or—alternatively—multiple robots with a single manipulator. We investigate mechanisms to jointly optimize multiple DMPs to perform one task in a coordinated fashion. The joint trajectory is built from the initial DMPs learned for a single manipulator, and its optimization must comply with task-specific constraints. We illustrate the application of our approach both in a simulated environment and in a simulated and real Baxter robot. I. INTRODUCTION In collaborative tasks, team members work together to accomplish a particular goal. Such teams may be composed of humans and robots, possibly more than one robot or more than one human. Collaboration towards the common goal requires coordinated actions between the members according to the stages of task completion and their skills. Usually, the skills of team members are complementary to each other to accomplish the required task [1] . It is crucial to clarify that the scope of the problems addressed in this paper does not encompass tasks that can be decomposed into sub-tasks, each capable of independent execution by individual robots, as is frequently encountered in multi-robot scenarios [2] . Our focus, instead, revolves around tasks necessitating the simultaneous execution of coordinated and synchronized movements, all the while accounting for environmental conditions and constraints. This scenario frequently arises in domains such as robotics [3] and gaming [4] . Several approaches exist that address the problem of coordinating multiple motions. For example, Stavridis et al [5] addresses the problem of assembling two parts where—instead of using a manipulator just to hold one part steady and the second to assemble the second part—both manipulators move to render the assembly easier. The movement of the two manipulators is described using a dynamical system; 1Ali Hani Kordia is with the SnT Centre, the University of Luxembourg, Luxembourg, and with INESC-ID and Instituto Superior Tcnico, University of Lisbon, Portugal; [ali.kordia@tecnico.ulisboa.pt](ali.kordia@tecnico.ulisboa.pt). 2Francisco S. Melo is with INESC-ID and with Instituto Superior Tcnico, University of Lisbon, Portugal; [fmelo@inesc-id.pt](fmelo@inesc-id.pt). the trajectories of the system are then adjusted to ensure that the task constraints are verified. Constraints include the desired position for the parts to be assembled, as well as the obstacles in the environment that should be avoided. There are two key differences in our work. First, an initial movement for both manipulators is provided upfront, which is then optimized using a hierarchical quadratic programming approach. In our work, we depart from a single-manipulator movement that is then used to construct the motion of both manipulators; additionally, following the approach in work [6], We have opted for Black-Box Optimization (BBO) over a quadratic programming optimizer, affording us the advantage of reduced model complexity. Unlike specific quadratic programming methods that necessitate linear or quadratic approximations of the objective function, BBO techniques operate directly on the black-box objective function w

FAQs

What problem does the paper address for collaborative manipulation tasks?

It addresses how multiple robots (or multiple manipulators) can produce simultaneous coordinated movements using a library of pre-learned primitive actions while satisfying task-specific constraints.

How is the proposed method built from DMPs?

It starts from DMPs learned for a single manipulator and extends them to construct and optimize coordinated executions for multiple manipulators or multiple robots under constraints.

Where is the approach evaluated?

The paper evaluates the framework in a simulated environment and also on a Baxter robot, including both simulated and real experiments.

Oct 3, 2026

Upload by anakgang17