Collaborative load transport using multiple robots
The use of a collection
of robots to execute a common task such as material transport or cooperative
assembly is becoming increasingly common as the costs of robotic hardware, processing
power and software are reduced. Using
multiple robots versus a single robot has the advantage of distributing a load
among several smaller and less expensive robots, and tighter control of the
internal force of the payload. In
addition, there may be increased dexterity in handling the payload (such as in
multi-finger manipulation), fault tolerance (defect of a subset of robots may
not completely derail the task), and reconfigurability
(the robots may be reconfigured to fit different distributed sensing and
actuation needs). Collaborative
transport of a load is also common in the biological world. Two ant species that are most proficient in
group transporting, Pheidologeton diversus
and Oecophylla smaragdina, form some of the largest perennial colonies. Indeed, ants have served as the motivation of
several mobile robot testbeds. The goal of this project is to develop fully
decentralized motion and control strategies for collaborative load transport.
We extend the centralized multi-robot motion and force control that we
previously developed to the decentralized case. As a simple initial case to
investigate, the robots are assumed rigidly attached to the load, and all
robots and the load are in a plane. We
also assume that there is no explicit communication of measured signals between
the robots, so the controller structure is fully decentralized. We adopt the move/squeeze decomposition
approach that we previously proposed and address the motion loop first without
considering the force and then study the force loop with motion induced force
as a disturbance. For the decentralized
motion control, we strengthen our previous result to semi-global exponential
stability. The decentralized force
control adds a perturbation term to the motion loop. Due to the robustness inherent in exponential
stability, the closed loop system remains stable. In the case that the desired motion of the
load is from rest to rest, both motion and force converge to the desired set points
exponentially.
Reports and Papers:
- J.T. Wen,
Acknowledgment
This work is supported in part by the National
Science Foundation under grant No. IIS-9820709.
Any opinions, findings, and conclusions
or recommendations expressed in this material are those of the author(s) and do
not necessarily reflect the views of the National Science Foundation.
This work is also supported in part by the Center for Automation Technologies
(CAT) under a block grant from the New York State Office of Science,
Technology, and Academic Research (NYSTAR).
Contact Information:
Office: CII 8213
Voice:
Fax:
Email: wen@cat.rpi.edu