Distributed object transportation on a desired path based on Constrain and Move strategy

https://doi.org/10.1016/j.robot.2004.08.006Get rights and content

Abstract

In this paper, a distributed strategy to move objects on different arbitrary paths in a 2D plane is proposed and analyzed. This algorithm which is based on Constrain and Move strategy [M.N. Ahmadabadi, E. Nakano, A Constrain and Move approach to distributed object manipulation, IEEE Trans. Robotics Automation 17 (2) (2001) 157], organizes the robots in two groups. The object manipulation task also is decomposed to two different tasks. The task given to one group is control of linear velocity and that assigned to the other group is control of angular velocity of the object. The independence of these tasks makes the design of the distributed architecture of the team possible. To calculate each robot's desired velocity, a simple method using Constrain and Move strategy and robot's local sensors is developed. To prevent small errors in the robot sensory system from affecting the system performance, limited compliance is assumed in robot arms. The basic behaviors of the robots are presented. Moreover, simulation results are given to verify the proposed strategy.

Introduction

Recent advances in micro and mini robot technologies have paved the way for using these robots for object manipulation. Considering these robots’ limited sensing and actuation capabilities, a large number of them is required for object manipulation. Regarding the mentioned limitations and also restrictions on computational and communication abilities of these robots, the control and coordination methods should be designed in such a way that their requirements don’t exceed the robots’ capabilities. In addition, the coordination and control protocols must be independent of the number of robots. Moreover, it should be possible to increase the system fault tolerance and to dynamically change the number of robots getting involved in the task. These requirements also exist in employing existing mobile robots for manipulation of large objects. Therefore, in this paper distributed object manipulation using a team of mobile robots with limited capabilities is studied.

The control of a robotic team is based on centralized, leader–follower or distributed methods (e.g. [2], [3], [4], [5], [6], [7], [8], [9], [10]). In [1], [11], it is discussed that, the central and the leader–follower schemes are expensive and vulnerable with respect to computation, communication and failure of the central unit or the leader. In contrast, in distributed systems, the task is distributed among the robots and each robot decides about its action by itself. Also, distributed systems could be made fault tolerant more easily. But, finding proper task distribution, which requires simpler coordination among the robots, is the main problem in a team of distributed object manipulating robots.

In distributed systems, force distribution is almost impossible. That is, determining the force, which should be applied by each robot, is too sophisticated. Therefore, using position-controlled robots is suggested.

In [6] and [12], Furniture Moving and BeRoSH are proposed. In these two strategies, the object is carried to the defined goal in a distributed manner but the path of object is not under control. In [1] and [11] the Constrain and Move strategy is proposed as a general distributed method. However, the object manipulation is limited to move the load along a straight line or to turn it around a defined point. Extending this strategy for moving the object on an arbitrary path considering the existing limitations in the robots’ sensing and actuation capabilities is the main goal of this paper.

In Section 2 a model of the system is presented. Constrain and Move strategy and the basic robot behavior and information system are reviewed in Section 3. Then the proposed algorithm is introduced. The results of the computer simulations are given in Section 5. A conclusion of this paper is given in Section 6.

Section snippets

A kinematic model of the robots

Each robot has a mobile base with an arm. Each arm is connected to the mobile base by a one D.O.F revolute joint and has two prismatic joints, see Fig. 1, Fig. 2. One of the prismatic joints is used for lifting the object and the other one is passive with two springs along the arm. The end-effector is connected to the arm by a ball joint.

The robots are controlled kinematically and it is assumed that the static and quasi-static forces dominate the mobile robot dynamics. Also, the arm is low

The Constrain and Move strategy [1]

Suppose that several robots lift the object and suspend it on its bottom face (Fig. 1). To turn the object around a point in a plane (Fig. 4a and b) the robots must constrain the object in X and Y directions and produce appropriate torque around the object's rotation axis. Also, to move the object on a straight line (Fig. 4c), the robots have to constrain the object in the desired path-perpendicular direction and push it along the path. In Constrain-Move strategy, these two tasks are

Extending Constrain and Move strategy for moving the object on a desired path

In Constrain and Move strategy, the constraint making robots immobilize a specific point of the object, point O in Fig. 7. That is, the constructed closure prevents point O from moving along X and Y axes. Consequently, the velocity-controlled robots can move point O at any desired velocity as they can make a force direction closure [1]. Now the main problem is to calculate the desired velocity of each robot–object contact point when the linear and angular velocities of the object are given.

For

Simulation results

The proposed algorithm is tested on a team of four simulated robots to move the object on some different types of paths. The model used is very close to the real model of robots used in [1]. In this model, the object is a 0.5 m × 0.5 m rectangle of 10 kg and 10 kg m2 mass and moment of inertia respectively. The mass center is placed at the geometrical center of the object. The position-controlling robots and orientation-controlling robot are located at (0.25, 0), (0, 0.25), (−0.25, −0.25) and (0,

Conclusion

In this paper, a distributed strategy to move objects by a robotic team was proposed and analyzed. This method which is based on Constrain and Move strategy, brings the position and orientation of the object under control. To achieve this goal, the robots are divided in two groups to do two different tasks. The task of one group is to control the object position and that of the other group is to control the object orientation. The success of the proposed algorithm in controlling the object

References (15)

  • M.N. Ahmadabadi et al.

    A Constrain and Move approach to distributed object manipulation

    IEEE Trans. Robotics Automation

    (2001)
  • D.J. Stilwell, J.S Bay, Toward the development of a material transport system using swarms of ant-like robots, in:...
  • K. Kosuge, et al., Motion control of multiple autonomous mobile robots handling a large object in coordination, in:...
  • Z.D. Wang, Y. Hirata, K. Kosuge, Control multiple mobile robots for object caging and manipulation, in: Proceedings of...
  • B.R. Donald

    Analyzing teams of cooperative mobile robots

  • D. Rus, B. Donald, J. Jenning, Moving furniture with teams of autonomous robots, in: Proceedings of the 1995 IEEE/RSJ...
  • E. Todt, G. Raush, R. Suarez, Analysis and classification of multiple robot coordination methods, in: Proceedings of...
There are more references available in the full text version of this article.

Cited by (17)

View all citing articles on Scopus
View full text