1. INTRODUCTION

Recently, times of emergency have necessitated rapid searches for victims in collapsed buildings. However, searches conducted by humans involve risks of secondary disasters. Therefore, it is expected that rescue robots search for victims before entry by humans [1].

Most existing rescue robots are controlled by teleoperation, because fully autonomous control may be dangerous in uncertain disaster environments. However, full teleoperation is difficult for a human operator, because a rescue robot has multiple degrees of freedom in general. Therefore, human support control system, “semi-autonomous control” in other words, is needed [2].

One of most critical incidents is falling down of the robot. Figure 1 shows a robot which falls down in the RoboCup Rescue Robot League competition. As shown in Figure 1, the fallen robot cannot run again without someone’s supports.

Figure 1

Fallen robot in RoboCup rescue competition.

Okada et al. [3] proposed a semi-autonomous control system to avoid the falling-down situation by using the normalized energy stability margin (S_NE) proposed by Hirose et al. [4]. However, the authors assume that the circumference of the robot contacts with the ground. Therefore, the system often cannot calculate the risk of the fall precisely, when the robot runs on a rough terrain, because the contact point between the robot and the ground may locate on the inner part of the robot.

In the existing works, the assumption about the contact points between the ground and the robot are necessary because S_NE is obtained from the position of the contact points and it is difficult to get the precise position of the contact point. However, the assumptions of existing studied do not often satisfy when the robot runs on a rough terrain.

In our previous work, we developed estimating system of the contact points by using a reconstructed 3D map from the point cloud data of a 3D depth sensor, such as KINECT [5]. Therefore, the purpose of this study is to develop a semi-autonomous system to avoid the falling-down situation. In our system, we use partial differentiation of the S_NE. The contact points are estimated by the system which is developed in our previous work.

2. CALCULATION METHOD OF NORMALIZED ENERGY STABILITY MARGIN

3. CONTROL METHOD OF SUB-CRAWLERS

To avoid the tumble situation, S_NE should be increased by moving the sub-crawlers, when the S_NE becomes almost zero. Therefore, the sub-crawlers should be moved so that

S_NE is a function of the contact points and the angle of the sub-crawlers. In this study, we assume that the contact points are known by the proposed estimation system described in Section 2 and S_NE is the function of the angle of the sub-crawlers only. Therefore, The definition of θ_YX = θ_FR, θ_FL, θ_RR, θ_RL is shown in Figure 5. The target angle of the sub-crawlers are set as Equation (4) in this study. By applying Equation (4) to Equation (3), we obtain the following Equation (5): Therefore, in the case of K > 0, the right side of Equation (5) is always positive, i.e., to avoid the tumble situation by satisfying Equation (2), the target angle of the sub-crawlers should be set as Equation (4).

Figure 5

Definition of θ_YX = θ_FR, θ_FL, θ_RR, θ_RL (Top view).

In the proposed control system, the robot is teleoperated by an operator. However, when the S_NE becomes almost zero, the target angles of the sub-crawlers are used as Equation (4) instead of the input by the operator to avoid the tumble situation. Figure 6 shows the control flow chart of the proposed control system.

Figure 6

Flow chart of the proposed control system.

4. EXPERIMENT

5. CONCLUSION

In this study, we propose a tumble-risk-assessment system, which calculates a normalized energy stability margin by using a 3D depth sensor. Moreover, a control system to avoid the tumble situation by moving the sub-crawlers automatically is proposed. We implemented the proposed systems to a real robot, carried out experiments, and confirmed the effectiveness of the proposed method.

In the situation where it is difficult for the existing to control the proposed method can increase S_NE by about three times.

The future works are to carry out the experiment on rough terrains and to develop a system not to become a dangerous situation by predicting the S_NE.

CONFLICTS OF INTEREST

The authors declare they have no conflicts of interest.

Authors Introduction

Dr. Noritaka Sato

He received Doctor of Engineering degree from The University of Electro-Communications in 2009. He was a Program-Specific Assistant Professor of Kyoto University in 2009–2011 and is an Assistant Professor of Nagoya Institute of Technology since 2011. His research interests include human robot interaction for rescue robots, rehabilitation-assist robots and industrial robots. He is a member of the IEEE, RSJ and SICE.

Mr. Kotaro Ohshima

He received Master of Engineering degree from Nagoya Institute of Technology in 2016. His research interests include control system, system integration and teleoperation system for rescue robots.

Prof. Yoshifumi Morita

He is a Professor in the Department of Electrical and Mechanical Engineering at Nagoya Institute of Technology. He received the PhD degree from Nagoya Institute of Technology in 1998. His current research interests are rehabilitation support robot/device and robot teaching.