VEA-II: Mobile Robot for Teleoperated and Autonomous Manipulation Missions

VEA-II: Mobile Robot for Teleoperated and Autonomous Manipulation Missions

Copyright © IFAC Intelligent Autonomous Vehicles, Madrid, Spain, 1998 VEA-II: MOBILE ROBOT FOR TELEOPERATED AND AUTONOMOUS MANIPULATION MISSIONS Jos...

1MB Sizes 0 Downloads 17 Views

Copyright © IFAC Intelligent Autonomous Vehicles, Madrid, Spain, 1998


Jose Manuel Ezkerra, Jokin Mujika, Jone Uribetxebarria, Luis Berasategi, Sabin Fernandez lKERLAN (Applied Research Centre) Robotics and Automation Dpt. P.o. Box 146, £-20500 Mondrag6n Spain Tel.: 34-43-771200 Fax: 34-43-796944 e-mail: [email protected]

Abstract: This communication describes developments completed within a project for research into the area of advanced robotics, specifically in the field of mobile robotics and remote handling, carried out in the Centre for Technological Research IKERLAN. The basic objective of the project is to design, manufacture and control a vehicle equipped with a manipulator arm, capable of operating both autonomously and remote-controlled, for its use in remote handling applications in indoors and outdoors environments. Copyright © 1998 IFA C

Keywords: Mobile Robot, Teleoperation, Navigation, Perception.



The prototype developed is divided into three clearly differentiated parts: a vehicle, a manipulator arm fitted to the vehicle and a control station. The control station serves as the interface with the operator, carries out planning not requiring real time characteristics and presents a visual monitoring of the missions in execution. The vehicle performs collision-free displacements between accessible points in its operating environment. These displacements can be made completely independently or remote-controlled by an operator physically and visually separated from the vehicle. The onboard arm carries out object manipulation tasks. Like the vehicle it can execute allocated tasks in an autonomous or remote-controlled mode. The figure 1 shows the vehicle with its fitted arm.

The intention in developing the experimental prototype was to contribute to the following specific technological objectives. a. Advanced techniques for locomotion To achieve this objective the approach was to seek novel electromechanical solutions which has to provide the vehicle with the capacity to make movements with both holonomic and non-holonomic restrictions. The system has to make it possible to test various locomotion configurations, which can range from configurations for directing the robot in an ornni-directional mode to configurations which require complex manoeuvres to access all the accessible points of its environment, turns on itself, etc. When developing this subsystem, priority was given to the factor that the robot must be able to manoeuvre on non-level surfaces, be capable of ascending and descending slopes, manoeuvring on bumpy terrain, etc. b. Advanced Perception Techniques In this aspect the approach was to study and develop perception techniques which allowed the system to act in tele-operated mode and in autonomous mode. In tele-operated mode, the system has to be controlled by an operator separated physically and visually from the working environment. The operator has to receive a large quantity of information about

Fig. 1. Mobile robot and manipulator. 605

the environment which enables execution of missions and ensures the integrity of the system. In autonomous mode, the system must have continuous information about its environment by means of sensors, to carry out the decisions taken automatically and continuous localisations.

reforming the plan generated depending on external events. 3. DEVELOPMENTS COMPLETED As already mentioned, the prototype developed IS divided into three clearly differentiated parts: a vehicle, a manipulator arm fitted to the vehicle and a control station. The operator introduces missions at the control station which carries out the planning for these missions. An example of an mission could be: 'Vehicle go to point A and Manipulator pick up type F object'. The result of the planning is the production of a plan of actions which are sent to the vehicle-arm unit which carries them out. The vehicle-arm unit is equipped with a series of sensors (ultrasonic, cameras, laser telemeter) which provide it with information on its working environment and enable to take action to meet unforeseen events (obstacles in its environment, changes in the slope of the terrain etc.).

c. Techniques for adaptation to the environment

Attainment of this objective needed study and development of sensor based navigation techniques. These techniques, applied to the manipulator, have to enable it to work in its environment by adapting to it and to carry out handling of objects or parts automatically. The vehicle has to execute collisionfree displacements, adapting the pre-programmed displacement to any unexpected obstacles, by avoiding these or effecting new movement plans. d. Remote handling techniques

As already mentioned, the vehicle-manipulator unit must be capable of operating in both autonomous and remote-controlled mode. For operating in remotecontrolled mode, remote handling techniques have been researched to enable an operator to direct the system without being physically and/or visually near the vehicle. To do this the information from the sensors fitted to the vehicle and in the manipulator has to be used. Conversely commands for remote controlled movement have to be generated so that the system operates to complete the required missions.

Each of the parts in which the prototype is divided is described below. 3.1. Control station

The control station is responsible for the control aspects which do not require a real time response. It effects the highest level programming and control tasks. Its role is to act as the interface for the user who, as mentioned above, introduces the missions to be completed by the vehicle-arm unit. It uses a planner which calculates the optimum paths to reach the accessible points of the environment. In the execution time of missions it at all times supervises their correct execution and offers the user a visual monitoring of the missions execution. The Central Station software has been implemented in a PC under Windows NT operating system. The ROB CAD tool has been used to perform the simulation and visual display of these mission executions.

e. Decision-making capacity of the vehiclemanipulator unit With a view to operating the system autonomously, the vehicle-manipulator unit must have a high decision-making capacity. With this objective study and development has been undertaken of techniques enabling the system to interpret high-level orders such as 'GO TO POINT', 'PICK UP OBJECT, 'BRING OBJECT TO OPERATOR', etc., by generating a plan of actions for completing missions autonomously, with the possibility of reacting and TCPI1P sockets


Monitoring Platform (UNIX) RS232

Central Station (NT)

On-Board Control (OS9)

Fig. 2. Control distribution. 606


3.2. Vehicle The mobile vehicle has a rectangular base and dimensions of 1400 x 800 mm. The locomotion system is based on four wheels, each of which acts independently in traction and direction. In this way the vehicle has been equipped with the possibility of adopting varying locomotion configurations, from configurations which are used for making omnidirectional movements (crab configuration) to configurations which give 'car-like' movements, reduced turns, turns on itself, etc. (figure 4).


...",. I


{o '...,.... "

° . . .., ,~


The vehicle is capable of working in external environments with terrain irregularities. To negotiate these possible irregularities the locomotion system has been designed so that each tractional-directional unit has suspension independent of the central chassis.

Fig. 3. Central Station Architecture. From a functional point of view, the most important modules of the Central station are presented in the figure 3.

The vehicle's trajectories control makes it possible to make displacements in any direction, and to make the manoeuvres necessary to take up position in any accessible configuration in its working environment.

The Central Station uses two processes (A and B processes in the figure) .The A process is subdivided into two threads: • A I: This thread is the user interface, involving the functions concerning direct execution of robot (vehicle + arm) commands ("DIRECT INTERACTION") and the functions needed to plan the missions assigned to the robot ("PLANIFICA nON"). There are three types of commands: control commands (changes of modes, stop, ...), data request commands (sensors data i.e.) and activity requirement commands (execution of plans, of movements, etc.). The "PLANIFICA nON" module is the heart of the system when it woks in autonomous mode.

With a view to the vehicle carrying out movements which are autonomous or controlled by an operator physically and visually separated from the working environment, it is necessary to equip the system with a unit of sensors which provide information on its environment. Information from these sensors is used by the vehicle's navigation module to make the collision-free displacements required to complete the missions allocated by the user in a totally autonomous way, or to achieve the remote presence necessary to complete remote-controlled operations reliably and safely. The sensors with which the vehicle is equipped are firstly a belt of 24 ultrasonic sensors providing information used by the controller to detect and avoid unforeseen obstacles and carry out tracking of surroundings reactively. It is also fitted with a telemeter laser which is used to localise the vehicle and to model free spaces in which the vehicle can operate. The said telemeter has been equipped with two orthogonal scanning systems. The set of points obtained by the telemeter is processed in such a way that a series of free planes in which the vehicle can operate is obtained. These planes do not have to be horizontal, for there are also sloped planes which the vehicle can access ..

• A2 : This thread (SEQUENCER) is normally stopped and waits a signal that notifies that a command or a plan has to be executed. It reads the corresponding data from a shared memory space and it transmits it to the SUPERVISION process. The B process is a unique thread called "SUPERVISION" that works as a DDE messages server. It sends the different commands to the robot using a radio link and receives from it its status. It supervises the correct functioning of the system. Finally it sends the result of the requested actions to the "SEQUENCER" and to the monitoring platform (ROBCAD).

Crab movement


Reduced gyro

Gyro around itself

Fig. 4. Locomotion configurations. 607

Finally, the vehicle has two cameras installed which are used as an element for stereoscopic remotepresence. All these sensors are capable of operating in indoors and outdoors environments.

information on the three forces and three moments. This information is used by the control to make adaptations of the trajectories to its environment by following contours with the end of the tool.


Sensor based navigation techniques studies and developments have been undertaken. These techniques enable the vehicle to operate in its environment by performing collision-free displacements, for which it adapts the preprogrammed displacement to possible unforeseen obstacles (fig. 5), by avoiding these or effecting planning movements by taking into account the successive models of the environment carried out by the sensorisation systems fitted to the vehicle. Fuzzy logic techniques have been used to develop these strategies. The fuzzy controller has been tuned by using genetic algorithms.



Fig. 5. Obstacle avoidance.


3.3. Manipulator arm

This communication describes developments completed within a project for research at IKERLAN. The development of this project allows to IKERLAN to deep in the mobile robotics field, in which it has been carrying out research activities since the earlier 90s.

The manipulator arm fitted to the vehicle is a PUMA 560 robot. It is fitted to the front part of the vehicle. A new trajectories control has been redesigned and developed for the said arm. ACT library has been used to develop the trajectories generator, which has been implemented in a VME based control system. This trajectories control enables the arm to make articular trajectories, and straight and circular trajectories relating to different reference systems and to perform sensor based movements.

The development of the project has demonstrated the functioning efficiency of the system, both in autonomous and teleoperated mode. In autonomous mode the system is able to execute complex missions with no need of any intervention of the operator. In teleoperated mode, a secure and simple control carried out by a physically and visually separated operator has been guaranteed. The demonstration of the efficacy of the system's different functionalities allows to IKERLAN to offer new projects involving the technologies developed.

A camera has been installed at the end of the arm enabling the user to carry out reliable remote control. This camera is also used to take images of the arm's field of handling, to subsequently process these, identify parts and automatic calculate the optimum position for handling the said parts. To obtain the vertical profile of the objects to be grasped, a laser light has been installed at the end of the arm. By making two perpendicular scans and analysing the position of the light in the images taken by the camera an approximated profile is obtained (fig. 6).

5. THANKS We thank the Basque Government as well as the CICYT (Comisi6n Interministerial de Ciencia y Tecnologia) for the help given for the carrying out of this project (ref. TAP95-0908-E and ref. TAP 960472).

Finally a torque and force sensor has been developed and fitted at the wrist end of the arm to provide


Min. distance


laser beam




Max. distance

Z = 11 + 12 = d / tg Cl + i / tg Cl

Fig. 6. Obtaining of the vertical distance to the vision plan. 608


Ezkerra 1.M., Mujika 1., Martinez A., Braunstingl R. "Design and development of a mobile autonomous robot at IKERLAN". ler Colloque international Productique Robotique du sud Europe Atlantique - Bourges - France (1995).

Betge-Brezetz S., Chatila R., Devy M. "Natural Scene Understanding for Mobile Robot Navigation". IEEE Robotics and Automation, voll, pp 730-737 (1994).

Fleury S., Herrb M. , Chatila R. "Design of a modular architecture for autonomous robots" . IEEE International Conference on Robotics and Automation, pp 3508-3513 (1994).

Braunstingl R., Mujika 1., Uribe J.P. "A wall following Robot with a Fuzzy Logic Controller Optimized by a Genetic Algorithm". Fuzzy Robot Control Competition Fuzzy-IEEE/IFESYokohama (1995).

Harrnut S., Huser 1. and Peters L "A Fuzzy system for Indoor Mobile Robot Navigation". FUZZIEEE/IFES'95, vol V, pp 71-76 (1995).

Braunstingl R., Sanz P., Ezkerra J.M: "Fuzzy logic wall following of a mobile robot based on the concept of general perception". International Conference on Advanced Robotics - ICAR Sant Feliu de Guixols (1995).

Laurette R., Alami R. "Supervision and control of the AMR intervetion robot". IEEE Robotics and Automation, (1991). Uribe 1.P., Urzelai 1. "Controlador fuzzy para la evitacon de obsticulos en un robot movil no holonomo". 7° Congreso Espaiiol sobre Tecnologias y logica Fuzzy (Tarragona 1997).

Campion G., Bastin G., D' Andrea-Novel B. "Structural properties and classification of kinematic and dynamic models of wheeled mobile robots". IEEE Transactions on Robotics and Automation VOL 12, NO 1, (February 1996).

Uribetxeberria J., Ezkerra 1.M., Mujika 1. "Planificador de trayectorias para un robot movil que evoluciona en entornos desconocidos". XVI Jomadas de AutomaticaIFAC (Donostia 1995).

De Luca A., Oriolo G. "Local Incremental Planning for Nonholonomic Mobile Robots". IEEE Robotics and Automation, vol 1, pp 104-110 (1994).

Urzelai 1., Uribe J.P., Ezkerra 1.M. "Fuzzy controller for wall-following with a non-holonomous mobile robot". Fuzzy-IEEE- Barcelona (1997).

Ezkerra J.M., Mujika 1., Martinez A., "Vehicul0 experimental autonomo de alta maniobrabilidad para entomos parcialmente estructurados". 3er Congreso de la Asociacion Espaiiola de Robotica, AER pp 217-224 (Noviembre 1993).

Yokokohji y., Ogawa A., Hasunuma H., Yoshikawa T. "Intelligent Teleoperation Systems".IEEE Robotics and Automation Vo!. 3 pp 510-515, (1993).