Figure 1. The seven-axis version of the DLR light-weight robot with integrated electronics and an early version of a three finger hand.
Résumé
Dans le prolongement de Rotex, premier robot télécommandé utilisé dans l'space, l'Institut de robotique et de dynamique des systèmes du DLR travaille à la conception d'une nouvelle génération de robots légers à capteurs multiples. Pourvus de mains articulées, ces robots peuvent être programmés à distance par des non-spécialistes pour exécuter des tâches définies dans un langage évolué. Des projets sont à l'étude pour exploiter au mieux les possibilités offertes par la robotique spatiale, par exemple pour réparer des satellites.
Contractors:
DASA (D), Tecnospazio (I), Jenoptik (D), OHB (D)
Funding:
German national programme, DARA (D) and ESA funding.
We made the first active step into space robotics with Rotex, the first remotely controlled robot in space. The provision of multi-sensory, local, shared autonomy, based on the use of a multisensory gripper and predictive graphics simulation, were crucial for the success of this historic project. The experiment flown on the US Space Shuttle, Columbia, in April 1993 could be operated under remote control of the astronauts on-board the Shuttle. It could be remotely controlled from the ground by man and machine, and by sensor-based off-line programming on-ground.
Presently the Institute's general goal is the design of a new generation of intelligent, multi-sensor light-weight robots, which can be operated by astronauts or from a ground station to perform maintenance tasks in space. These robots are based on powerful concepts of remote control and man-machine-interfaces which dispense with the need for a robotics expert at the ground-station.
Following a philosophy of embedding the electronics in the mechanical components, a modular design has been adopted for the new light-weight robot developed by DLR (Figure 1). Each module consists of a joint, a (carbon fibre grid) link-structure element and its electronics. An inductive torque measurement and feedback system is an integral part of the gearing system. Double planetary gearing is used to obtain a reduction ratio of 1:600 from an extremely small and light joint, which delivers a torque of 120 Nm yet weighs just 1 kg.
Figure 1. The seven-axis version of the DLR light-weight robot with integrated electronics and an early version of a three finger hand.
The design objective is to achieve a light-weight robot with a maximum load to weight ratio of about 1:2, whose manipulator system has seven degrees-of-freedom and a good dynamic performance. Total system weight should be about 15 kg, and there should be no bulky external wiring nor separate electronics cabinet, often found in industrial robots.
There are many possible configurations for a seven-axis manipulator, including options that allow the robot to be folded up into a very small volume for storage. For many tasks to be performed in a laboratory environment, a two-finger gripper is adequate. With this in mind, we have revised the design of the Rotex multi-sensor gripper used in ESTEC's robotics testbed. The Rotex gripper with its sixteen sensors (two different types of force-torque sensors, 9 laser range finders, tactile arrays and stereo cameras) and more than a thousand electronic components appears to have been the most complex robot gripper built so far. The more recent ESA version has only one force-torque sensor with 6 degrees-of-freedom but it contains a number of improvements (Figure 2).
Figure 2. An improved and simplified ESA version of a multi-sensory gripper (drawing).
To perform more complex manipulations, future space robots will need articulated multi-fingered hands; DLR's design employs a concept of artificial muscle, which combines planetary roller screw gearing (used in the Rotex gripper) with tiny motors to produce small, powerful, highly controllable actuators, which seem to be superior to hydraulic and pneumatic devices.
A four-fingered robotic hand is under construction using this technology. Each finger has 3 active degrees-of-freedom and contains 28 sensors (position, force, tactile).
Within the framework of DARA's Marco project, we are presently developing a system for remote control of future space robots and automata, which has a modular architecture and which can be embedded into a very general mission control system. Its core is a highly interactive, task-oriented, remote programming system based on learning by example in a virtual environment with four hierarchical levels. This would allow a non-specialist to control a robotic system by simply pointing with a three-dimensional cursor (or any other virtual reality interface) to an object and selecting the operation it must perform. An advanced man-machine interface, which makes the operator oblivious of the underlying details the robot control system, is provided.
The foundations of this concept, which combines remote sensing and remote control, were already laid in Rotex, and thus the new system can handle any arbitrary combination of off-line programming and on-line remote control using feedback from sensors (Figure 3).
Figure 3. The new telerobotic station at DLR.
The control techniques we have developed have found immediate application in the feasibility study of an Experimental Servicing Satellite (ESS) funded by DARA1. This is a free-flying robotic spacecraft, which can approach, inspect and repair a malfunctioning satellite in orbit. A specific example is TVSat-1, where one solar panel failed to unfold after launch. To support this project, a special capture tool has been developed for ESS, which uses real time video images, six laser range finders and force sensors to autonomously attach itself to the apogee motor of the target satellite (Figure 4). In addition, our laboratory prototype system, based on two industrial robot systems, can simulate the dynamic interaction between robot and satellite.
Figure 4. Capture tool for the ESS project
The Japanese ETS VII project, a free-flying robotics project in which DLR is participating, provides another example, in which a remotely controlled robot can be piloted without colliding with its environment and a global model of its environment is automatically updated.
We believe that the appropriate technology is available and the time is ripe for extensive use of robots in space, and what we now need are political decisions. Rotex - the first remotely controlled robot in space - has laid the foundation for payload servicing inside the pressurised module of the International Space Station. In our opinion, operational systems (and possibly mobile systems) of this type which can manipulate probes and be controlled from ground without the continuous presence of a robotics expert, are now definitely feasible. In addition to a much improved Rotex gripper, ultra light-weight robots and articulated, multi-fingered, hands will become available in the next few years. Experimental free-flying, remote controlled, robotic systems are being developed in several countries, as pre-cursors to operational systems (e.g. the Japanese ETS VII project), and we are convinced that, even in the mid-term perspective, the urgent application of robotics in space will bring substantial cost savings.
1. Settelmeyer E., et al, The Experimental Servicing Satellite, (in this issue of Preparing for the Future).
Preparing for the Future Vol. 7 No. 2