A reinforcement learning approach to fail-safe design for multiple space robots—cooperation mechanism without communication and negotiation schemes

This paper explores a fail-safe design for multiple space robots, which enables robots to complete given tasks even when they can no longer be controlled due to a communication accident or negotiation problem. As the first step towards this goal, we propose new reinforcement learning methods that help robots avoid deadlock situations in addition to improving the degree of task completion without communications via ground stations or negotiations with other robots. Through intensive simulations on a truss construction task, we found that our reinforcement learning methods have great potential to contribute towards fail-safe design for multiple space robots in the above case. Furthermore, the simulations revealed the following detailed implications: (i) the first several planned behaviors must not be reinforced with negative rewards even in deadlock situations in order to derive cooperation among multiple robots, (ii) a certain amount of positive rewards added into negative rewards in deadlock situations contributes to reducing the computational cost of finding behavior plans for task completion, and (iii) an appropriate balance between positive and negative rewards in deadlock situations is indispensable for finding good behavior plans at a small computational cost.     

A micro-rover navigation and control system for autonomous planetary exploration

This paper describes an end-to-end control system for autonomous navigation of a small vehicle at a remote place, e.g. in space for planetary exploration. Due to a realistic background of this study the proposed method has to deal with limited knowledge about the environment as well as limited system resources and operational boundary conditions, especially a very large time delay in the communication between the ground control station and the space segment. To overcome these constraints the remote system has to act in a very autonomous way. Ground support minimizes the computational load of the remote system. High-level information interchange reduces the communication bandwidth requirements.     

Decentralized Traffic Control for Non-Holonomic Flexible Automated Guided Vehicles in Industrial Environments

This paper deals with the problem of coordinating flexible automated guided vehicles (AGVs) in real manufacturing systems. The problem consists of ensuring safe and successful task execution while several AGVs operate as a distributed transportation system in real industrial environments. The proposed solution combines different decentralized techniques to increase the flexibility and scalability of the multirobot system. The coordination is addressed by dividing the problem into path planning, obstacle avoidance and traffic control problems. The path planning method takes into account the location of mates for replanning the routes. The obstacle avoidance technique considers the kinematic constraints of the platform for reactive motion control. The traffic control approach makes use of a decentralized control policy that takes into account the capabilities of vehicles. By combining all these techniques and configuring the system properly, we present the successful development of a distributed transportation system composed of a team of flexible AGVs. The proposed solution has been validated using both a set of custom-modified AGVs operating in a real factory and a simulation of several AGVs operating in a virtual scenario.     

CRABOT: A Biomimetic Burrowing Robot Designed for Underground Chemical Source Location

This paper describes the biomimetic inspiration and development of a burrowing robot called CRABOT. This robot is one part of a project to investigate novel techniques for locating the source of volatile chemicals released underground. A range of chemicals, including petrol, kerosene and diesel fuel, are commonly stored in underground tanks or transported by buried pipelines. All of these structures are prone to corrosion and the development of leaks. Currently, finding the source of underground chemical leaks is a difficult task. As part of an automated system to locate chemical leaks a prototype burrowing robot has been developed modeled on the mole crab Emerita talpoida. The current version of this robot functions in a 'semi-submersed' configuration that allows the operation of the burrowing mechanism to be easily observed. Later versions will burrow completely below the surface. This paper provides details of the design and performance of the prototype burrowing robot.     

Cooperative task excecution of a search and rescue mission by a multi-robot team

Rescue robots have proved to be an extremely useful work partner for urban search and rescue (USAR) missions. Human rescuers who carry out these missions frequently enter dangerous zones to search for survivors; however, due to the unstable nature of collapsed buildings or objects, their lives may also be threatened. For this reason, in order to reduce life-threatening risks, rescue robots are deployed to carry out the job instead. Rescuers can now operate the robots at a safe place while the missions are carried out. When the robots have gathered enough information about the location of the victims and data about their physical conditions, rescuers can then enter the disaster site with enough knowledge to avoid harm and rescue the victims in the shortest time possible. In this paper, we introduce examples of 'effective multiple robot cooperative activities' and 'a study of the number of robots and operators in a multi-robot team' from our experiences gained from participating in RoboCup Rescue competitions.     

Experimental realization of dynamic walking for a human-riding biped robot,HUBO FX-1

This paper describes a control strategy of the stable walking for the human-riding biped robot, HUBO FX-1. HUBO FX-1 largely consists of two legs with 12 d.o.f., a pelvis and a cockpit. A normal adult can easily ride on HUBO FX-1 by means of a foothold, and can change the walking direction and speed continuously through the use of a joystick. Principally, this kind of robot must be able to carry a payload of at least 100 kg in order to carry a person easily. A sufficient payload can be accomplished by two ways. The first is through the choice of a highly efficient actuator. The second is through weight reduction of the robot body frames. As an efficient actuator, a high-power AC servo motor and a backlash-free harmonic drive reduction gear were utilized. Furthermore, the thickness and the size of the aluminum body frames were sufficiently reduced so that the weight of HUBO FX-1 is light enough. The disadvantage of the weight reduction is that HUBO FX-1 was not able to walk stably due to structural vibrations, as the body structures become more flexible due to this procedure. This problem was solved through the use of a simple theoretical model and a vibration reduction controller based on sensory feedback. In order to endow the robot with a stable biped walking capability, a standard walking pattern and online controllers based on the real-time sensory feedback were designed. Finally, the performance of the real-time balance control strategy was experimentally verified and stable dynamic walking of the human-riding biped robot, HUBO FX-1, carrying one passenger was realized.     

Study of Super-Mechano Colony: concept and basic experimental set-up

In our COE/TITech project, we have been studying a heterogeneous robotic system of decentralized autonomous agents with a leader agent, known as the 'Super-Mechano Colony' (SMC), which has characteristics of both the hierarchical and distributed control systems. This paper explains the basic concept and a physical test-bed of a SMC and discusses one example, specifically planetary rovers, for SMC. In order to provide a test-bed for the basic architecture, a homogeneous colony of 12 child-machines and one mother-ship were designed and built. Each child-machine of the colony has an octagonal body mounted on two independent axle wheels and each robot is equipped with a manipulator arm. Each robot has 5 actuated d.o.f.; specifically, two motors rotating the independent wheels of the body, one motor for opening and closing the gripper, one motor for lifting and lowering the arm, and one motor for rotating the body. While the child-machines are grasping the stud attached around the mother-ship, each child-machine can charge its own battery from the charger mounted on the mother-ship.     

Walking Pattern Generation for a Biped Walking Robot Using Convolution Sum

This paper describes a novel walking pattern generation method for a biped humanoid robot using a convolution sum. For a biped walking model, a single mass inverted pendulum model is generally used and the zero moment point (ZMP) equation is described by a decoupled linear differential equation. As a walking pattern generation method for the robot model, a novel method using a convolution sum is proposed in this paper. From the viewpoint of the linear system response, walking pattern generation can be regarded as a convolution of an arbitrary reference ZMP and the walking pattern for an impulse reference ZMP. For the calculation of convolution, the walking pattern for an impulse reference ZMP is first derived from the analytic walking pattern for a step reference ZMP. The convolution sum is then derived in two recursive forms, which can be applied online and offline, respectively. The proposed algorithm requires low computation power, since the walking pattern equation is composed of a recursive form. As the algorithm is expressed in analytic form, it is not necessary to solve optimization problems or calculate the fast Fourier transform, contrary to previous approaches. A computer simulation of walking demonstrates that the proposed algorithm yields excellent accuracy compared to the preview control method — one of the most highly regarded walking pattern generation methods. In addition, the application on the multi-point mass model is shown with the computer simulation.