Elsevier

Robotics and Autonomous Systems

Volume 110, December 2018, Pages 92-101
Robotics and Autonomous Systems

Mechanism allowing large-force application by a mobile robot, and development of ARODA

https://doi.org/10.1016/j.robot.2018.09.005Get rights and content

Highlights

  • A mechanism for a mobile robot to apply a large force without falling is proposed.

  • The mobile robot with the proposed mechanism named ARODA is developed.

  • The ARODA successfully tilted a heavy object while searching the unknown parameters.

Abstract

This study proposes a mechanism and a methodology for large force input to the environment by a mobile robot. To determine the limits on the force that a mobile robot can apply to a target object, we analyzed the forces between the robot, ground, and object, and the frictional-force limits between any two of these three bodies. To prevent the mobile robot from falling during the large-force application, the manipulator is connected to the robot via a passive rotational joint. This mechanism enables the mobile robot to search the environmental parameters. A new mobile robot fitted with the proposed mechanism, named ARODA, was developed. In a validation experiment, the developed mobile robot successfully tilted a relatively large and heavy object while searching the environmental parameters (the frictional coefficients of the floor and object and the size of the object). Equipped with the proposed mechanism, the mobile robot refrained from falling while applying a large force to the object by trial and error.

Introduction

A target object manipulated by a robot is often assumed to be firmly grasped by the robot’s end-effector. The place-to-place transfer of objects by mobile robots, called pick-and-place, is a widely used technique in the industrial robotics field. However, if the robot is designed to explore unknown environments and manipulate unknown objects, the object may be ungraspable because of its size or the robot cannot reach to the location where the robot can grasp the object in some cases. For example, it is difficult to apply a large scale robot with a large end-effector to domestic environments because of space, but sometimes the robot may be required to move relatively large furniture to replace it or relatively large packages to carry it to the appropriate location. It is hard to imagine that a domestic-size mobile robot conducts pick-and-place to such kind of large targets. As another example, mobile robots are required to enter a restricted space in disaster sites removing obstacles. The pick-and-place approach may be unworkable in this case too because obstacles are not always small enough to pick up. Instead, the robot must be able to manipulate the object without grasping it.

As an alternative to pick-and-place, a robot may slide an object by pushing it [1], [2], [3], [4], [5], [6], [7]. The planning and stable execution of pushing manipulations have often been studied in wheeled mobile robots [1], [2], [3]. Recently, control methods for the pushing manipulation of legged humanoid robots have been proposed [4], [5], [6], [7], and these methods extend the control method based on the zero moment point (ZMP). Whereas only sliding an object by pushing is considered in these works, other approaches consider the rotation of an object by a single or multiple robots [8], [9]. Collectively, these approaches are known as nonprehensile or grasp-less manipulations [10], [11], [12], [13]. Nonprehensile manipulation requires a lower force application than grasping manipulation, because it transfers the gravitational force acting on the object to the environment. Furthermore, nonprehensile manipulation can maneuver an object placed in very narrow or obstructed spaces, which typify the abovementioned domestic and disaster areas. Therefore, nonprehensile manipulation is particularly useful for manipulating a relatively large object by a small mobile robot. However, nonprehensile manipulation of a relatively heavy object is more difficult for small mobile robots than for other kinds of robot, because the robot is easily toppled or slipped by the large reaction force. When the mobile robot is required to tilt or pivot the object without grasping, the toppling risk is especially significant.

To prevent slippage or toppling, one must consider the frictional force between the object and the end-effector of the robot [14]. In nonprehensile manipulation tasks, the robot must often support the object at one or more frictional contact points in a small area without slippage. Furthermore, in non-grasping manipulation tasks, the frictional force between the mobile robot and the ground is also important. Therefore, to ensure reliable planning of nonprehensile manipulation by a mobile robot, the pushing capability of the robot must be defined taking into account the friction between the robot and the ground, in addition to the friction between the robot and the object.

However, even when the pushing capability is clarified, environmental uncertainties in the object and the ground cause errors in the manipulation result. To reduce these uncertainties, the robot must estimate the actual parameters by trial and error. The estimation of frictional property of the object by trial and error has been studied [15], [16], [17]. The estimation of the shape and center-of-mass location of the object by trial have also studied [18]. In any case of trial-and-error manipulation, toppling is a fatal problem for a roving mobile robot, because once fallen, the robot cannot easily restore its initial state. Therefore, the rotational forces applied to the robot through the contact point must be carefully controlled. However, for a small mobile robot manipulating a relatively large and heavy object, even small errors and disturbances impart a large reactive force on the robot.

Therefore, in the present research, a linear actuator is connected to the mobile robot through a passive rotational joint. This mechanism avoids toppling of the robot by an unexpected reactive force, even when the robot imparts a large force to an object. The new type of mobile robot fitted with this mechanism realize the manipulation of the object by trial and error in uncertain environments. Furthermore, to clarify the pushing capacity of the mobile robot, we also calculate the force limitation of the robot taking into account both of the limits of the robot-ground and robot-object frictional forces, and the rotational force that can be applied to the robot during the nonprehensile manipulation. We show that the force application to the object is more directionally restricted for a mobile robot with the proposed mechanism than for a mobile robot with a conventional manipulator, but the maximum applicable force is the same in both cases.

The remainder of this paper is organized as follows. Section 2 analyzes the pushing capability of a mobile robot, accounting for the frictional limitations of the robot-object and robot-ground interactions. Section 3 explains our proposed mechanism, which uses a passive joint to avoid falling of the mobile robot. A mobile robot fitted with the proposed mechanism, named the Autonomous Robot with One-dimensional Actuator (ARODA), is introduced in Section 4. In Section 5, the proposed mechanism is validated in a series of experiments. The developed robot remained upright when unexpected slippages occurred on the object or the ground, and identified the uncertain parameters (the frictional coefficients and the size of the object) by trial and error.

To simplify the analysis, we here limit the motions and forces to a two-dimensional plane. In most of the practical manipulation task, the forces applied to the robot is not restricted in the two-dimensional plane, but the analysis and the proposed mechanism are extendible to three-dimensional space. We discuss the problem and solution for the extension of the discussion in this paper to three-dimensional space in the final section.

Section snippets

Pushing capacity of a mobile robot

This section analyzes the maximum pushing force that a mobile robot can apply to an object. A mobile robot pushes an object in front of its body by direct contact (see Fig. 1). The floor and object surfaces contacted by the robot are assumed to be flat. Panels (a) and (b) of Fig. 1 illustrate the mechanisms of hand-pushing by a legged humanoid robot and manipulator-pushing by a wheeled robot, respectively. In the following discussion, the force balance is analyzed in the spatial frame fixed on

Anti-falling mechanism for the mobile robot and its limitation of force direction

When deriving the region of forces that the robot can apply to the object, we must consider the falling risk of the robot under a rotating force. For simplicity, we assume that frictional contact occurs between the robot and object. As the robot begins to fall, the points at which the moment acts on the robot are balanced statically outside the region of contact points with the ground, as shown in Fig. 5. Such a point is called a foot-rotation indicator (FRI) [24]. When the FRI is outside the

Development of a mobile robot with a passive joint

In our mobile robot Autonomous Robot with One-dimensional Actuator (ARODA), a manipulator is connected by a passive rotational joint, allowing large force exertion on an object. The ARODA is photographed in Fig. 8(a). The ARODA can move in any direction on its three omni-wheels. A manipulator comprising an end-effector and a linear actuator (FA-400-L-12-30, Firgelli Automations, Ferndale, WA, USA) is connected to a plate that contacts the ground via a passive rotational joint. The passive joint

Experiment

To confirm that the robot is not toppled by an unexpected external force, and can identify the environmental parameters through its sensors, we designed a series of trial-and-error experiments. In these tasks, the robot was required to apply a large force to tilt a heavy object. The experiments tested three situations; slippage between the object and the robot, slippage between the ground and the robot, and successful tilting of the object. In each situation, the frictional coefficients of the

Conclusions

This paper proposes a new design with a passive joint for a mobile robot, which allows the large-force manipulation of objects while preventing the robot from toppling. We first clarified the limitations on the forces that can be applied to the object (beyond which the robot will fall) by analyzing the frictional forces between the robot and object, and between the robot and ground. Next, we discussed the advantages and disadvantages of installing a passive joint in our proposed mobile robot.

Acknowledgment

A part of this study is the result of “HRD for Fukushima Daiichi Decommissioning based on Robotics and Nuclide Analysis” carried out under the Center of World Intelligence Project for Nuclear S&T and Human Resource Development by the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Shouhei Shirafuji received his Ph.D. degree in Information Science from Osaka University, Japan, in 2014. He was a JSPS Research Fellow from October 2014 to March 2015. Since 2015, he has been a Post-Doctoral Researcher with the Research into Artifacts, Center for Engineering, The University of Tokyo, Japan. His main research interests include mechanical design, robotics, and bio-mechanics.

References (25)

  • VukobratovićM. et al.

    On the stability of anthropomorphic systems

    Math. Biosci.

    (1972)
  • K. Lynch, The mechanics of fine manipulation by pushing, in: Proceedings of the IEEE International Conference on...
  • LynchK. et al.

    Stable pushing: Mechanics, controllability, and planning

    Int. J. Robot. Res.

    (1996)
  • MasonM.

    Mechanics of Robotic Manipulation

    (2001)
  • K. Harada, S. Kajita, K. Kaneko, H. Hirukawa, Pushing manipulation by humanoid considering two-kinds of ZMPs, in:...
  • T. Takubo, K. Inoue, K. Sakata, Y. Mae, T. Arai, Mobile manipulation of humanoid robots-control method for com position...
  • TakuboT. et al.

    Pushing an object considering the hand reflect forces by humanoid robot in dynamic walking

  • M. Murooka, S. Nozawa, Y. Kakiuchi, K. Okada, M. Inaba, Whole-body pushing manipulation with contact posture planning...
  • YoshidaE. et al.

    Pivoting based manipulation by a humanoid robot

    Auton. Robots

    (2010)
  • M. Murooka, S. Noda, S. Nozawa, Y. Kakiuchi, K. Okada, M. Inaba, Manipulation strategy decision and execution based on...
  • Y. Aiyama, M. Inaba, H. Inoue, Pivoting: A new method of graspless manipulation of object by robot fingers, in:...
  • Y. Maeda, H. Kijimoto, Y. Aiyama, T. Arai, Planning of graspless manipulation by multiple robot fingers, in:...
  • Cited by (2)

    Shouhei Shirafuji received his Ph.D. degree in Information Science from Osaka University, Japan, in 2014. He was a JSPS Research Fellow from October 2014 to March 2015. Since 2015, he has been a Post-Doctoral Researcher with the Research into Artifacts, Center for Engineering, The University of Tokyo, Japan. His main research interests include mechanical design, robotics, and bio-mechanics.

    Yuri Terada received his B.E. (2016) degrees from the Department of Precision Engineering, Graduate School of Engineering, the University of Tokyo. His research interests include mechanical design of a mobile robot.

    Tatsuma Ito received his B.E. (2017) degrees from the Department of Precision Engineering, Graduate School of Engineering, the University of Tokyo. He is pursuing the master degree in Precision Engineering at the University of Tokyo. His research interests include manipulation planning of multiple mobile robots.

    Professor Jun OTA is a Professor at Research into Artifacts, Center for Engineering (RACE), the University of Tokyo. He received B.E., M.E. and Ph.D. degrees from the Faculty of Engineering, the University of Tokyo in 1987, 1989 and 1994 respectively. From 1989 to 1991, he joined Nippon Steel Cooperation. In 1991, he was a Research Associate of the University of Tokyo. In 1994, he became a Lecturer. In 1996, he became an Associate Professor. From April 2009, he became a Professor at Graduate School of Engineering, the University of Tokyo. From June 2009, he became a Professor at Research into Artifacts, Center for Engineering (RACE), the University of Tokyo. From 2015, he is a guest professor of South China University of Technology. From 1996 to 1997, he was a Visiting Scholar at Stanford University. He received RSJ (the Robotics Society of Japan) Fellow in 2016. His research interests are multi-agent robot systems, embodied-brain systems science, design support for large-scale production/material handling systems, human behavior analysis and support.

    View full text