Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Direct Manned Interaction, Automation, Robotics, and Telescience MANNED INTERACTION The presence of humans allows for interaction with experiments and repair of malfunctioning equipment. At the same time, human presence degrades the quality of the microgravity environment; for that reason, it often is desirable to observe experiments and perform many tasks without direct human involvement. Where experiments on a CDSF or other free-flyer are concerned, it becomes not only desirable but mandatory to rely on automation, robotics, and telescience. The following section explores the unique value of having humans in space at this stage of our understanding of the behavior of materials and processes in space and assesses the state of the art in A&R and telescience. In a normal terrestrial setting, the fluid, material, and life sciences are researched by experimenters who are trained observers, astute to the appearance of unusual occurrences or unpredicted behavior. The situation in microgravity research, ideally, should be no different: the trained scientist should remain in close contact with his or her experiment. However, the rigor and cost of spaceflight is severely limiting to a human presence, and the practical conduct of science in space must compensate for this limitation. The short history of microgravity research has shown that most experiments benefit greatly from human presence, but, as mentioned earlier, the chief drawback is the accompanying and usually unavoidable degradation of the microgravity environment. The solution to the problem of how to involve researchers in microgravity research without accepting the interference of their associated perturbations or accelerations is to establish effective, near real-time telecommunication and teleoperation links between the terrestrial and orbital laboratories. Teleoperation combined with limited direct manned interaction may indeed be the best approach for many applications. This approach was used as early as the Skylab missions, in which astronauts could describe microgravity phenomena as they occurred to scientists on the ground, and on recent Spacelab 41
flights, in which mission specialists carried out critical on-orbit repairs on malfunctioning automated microgravity equipment, thereby rescuing several experiments from total failure. In the future, entire space experiments could be teleobserved and/or teleoperated from the ground. NASA's plans for microgravity R&D in the 1990s include use of the U.S. Microgravity Laboratory (USML), the U.S. Microgravity Payload (USMP), as well as secondary payloads such as middeck lockers, "Get-Away-Specials," attached payloads, and so on. These payloads and locations vary considerably in their ability to support up-linking and down-linking to Earth-based scientists, but each experimental mode is an opportunity for NASA and the microgravity community to further develop telescience capabilities. When the Space Station era starts in the late 1990s, there will be an opportunity for truly long-term, nearly continuous microgravity exposures, combined with the desired manned presence, and augmented with more advanced telescience. AUTOMATION, ROBOTICS, AND TELESCIENCE Whether performed by a human, a machine, or some combination of the two, most microgravity experiments still require close monitoring and control, over a period ranging from seconds to weeks, of many variables, all of which would obviously differ in number and kind for different experiments. Some form of automation has been used from the outset in such experiments, such as in generating carefully planned inputs to the experiments and measuring and recording responses. Ideally the principal investigator would like to be in space to make visual observations, especially of phenomena that are not easily captured by instruments and automation, and to reconfigure the experiment during the mission or to make repairs in case of failure. Delegating these functions to Space Shuttle mission specialists has generally worked well, and such "human-tending" has indeed saved several experiments. The salient question is to what extent in the 1992-1997 time frame the mission specialist can be aided or replaced by automation, robotics, or teleoperation, to make feasible the use of periodically human-tended or unmanned free-flyers as experimental facilities. Automation and robotics (A&R) is far from a stagnant field, and many recent advances have been demonstrated in the laboratory and in industrial applications. An example is computer visual and tactile recognition and performance of simple assembly and disassembly tasks at speeds and accuracies an order of magnitude greater than those attainable through human performance. Another example is computer-based intelligent decision-making (in which there is a well-established knowledge base). NASA microgravity research automation requirements are different from those of production-line automation, in which conditions are predictable, easily controllable, and repetitive. Microgravity research sensing and control needs are typically one-of-a-kind, and full automation would have 42
to be tailored to the individual experiment. Even though computer hardware configurations might be shared among many researchers, the software, sensing, and control automation hardware may have to be unique and tailored to each individual experiment. There are so many unpredictable aspects of most microgravity experiments today that providing fully autonomous operation (i.e., no human observation or intervention during the flight mission) is often too much to ask of automation and robotics. During at least the early stages of experimental work, the appropriate responses for all of the contingencies cannot be anticipated and programmed. This does not mean that the only alternative is experiment tending by a person who is physically present, with all of the associated costs and overhead constraints. An alternative that holds much promise for microgravity research is telescience or teleoperation, wherein the principal investigator observes the experiment from the ground (or a mission specialist does so from another orbiting vehicle). Using video and other modes for sensing, communications, and display, the investigator reprograms the on-board computer and/or moves a joystick or multiaxis hand device to control various actuators on the experiment. Such operator control devices can be simple built-in knobs or switches or multiaxis handles that can be positioned to control in-space manipulators to perform minor modifications to the experiment or to repair the apparatus when it fails. NASA has had an active program in automation and robotics for many years. Public Law 98-371, which took effect in 1984, gave it a further boost, committing 10 percent of the Space Station budget to A&R in one form or another. Perhaps even more significant is the development over three decades of teleoperated submarines for use in the deep ocean by the oil industry and the Navy and development of similar devices for nuclear "hot laboratories." There is much accumulated experience in performing remote viewing and manipulation (telescience) tasks in the laboratory and in the two application areas mentioned above. Human operators, given modest training and current state-of-the-art video devices using remotely controlled pan, tilt, and zoom functions, and current state-of-the-art five or six degree-of-freedom telemanipulators, can easily do requisite observation and manipulation to perform simple assemblies, adjustments, and repairs. There can be difficulties with depth perception, but stereopsis and multicamera techniques are being developed. Continual improvements in fineness of dexterity are being made as well, including touch and proximity sensors and displays, and operator adjustment of the impedance (mechanical stiffness and viscosity) to make the manipulation either compliant and gentle or stiff and precise, as appropriate to the task. Special problems have been posed by the existence of communication time delays in teleoperation control loops, whether caused by the finite speed of light or by the multiple signal processing delays in computers of the Tracking and Data Relay Satellite System (TDRSS) or ground stations. In either case, the result is two to six second round-trip delays that
force the human operator to repetitively make small movements and wait for confirming feedback, thereby making tasks take two to ten times longer than they would with no delay, or five to 25 times longer than they would if done by hands. One way around this problem is to use "supervisory control" or "telerobotics" systems, whereby the human operator sends packets of instructions to a remote computer/robot (telerobot) to perform a task segment. The telerobot uses its own tactile or optical sensors as references ("move in direction x until touch, then back off, open jaws and move up and grasp ..."), that is, the control loop is closed locally, with no time delay, and thus the whole operation can be accelerated and made more reliable. Such telerobots, which can also fall back on the more primitive direct master-slave teleoperation, are being developed experimentally by the Jet Propulsion Laboratory and the Marshall Space Flight Center. The two-arm, one-leg Flight Telerobotic Servicer, which is being designed for use on the Space Station Freedom, is being developed at the Goddard Space Flight Center. NASA is also developing miniature displays to be worn on the operator's head that would send control signals to point the video camera in the same direction as the operator's eyes, thus giving him or her a sense of being there ("telepresence"). Most likely to be available for use in space in the near term, say prior to 1995, are teleoperated video cameras that pan, tilt, and zoom, and single manipulator arms that are controlled in direct master-slave fashion. Such techniques will allow relatively slow control movements by the human operator, which are nevertheless more satisfactory than having no ability to remotely human-tend the experiment, and in most cases probably are tolerable. In fact, these time delays can be ameliorated through use of computer-based systems that take the operator's control inputs, model the geometry of the task and kinematics of the manipulator, and overlay on the delayed video an undelayed stick figure model of where the hand or end point of the manipulator is predicted to be, thus speeding up the operator's ability to make confident moves. Another form of computer automation that has seen rapid progress recently is one that provides the ability to process a variety of signals, make comparisons to updated process models as well as an a priori data base, and provide early warning of abnormalities or failures. Such computations could be done in the space vehicle or on the ground. Many other expert systems and computer-based decision aids are becoming available, with progress driven in part by the DOD strategic computing program. CONCLUSIONS Technology for teleoperation and computer-assisted decision-making has not yet been used to a great extent in the designs of microgravity experiments. The microgravity researchers on the committee stressed the current importance of human oversight of experiments, whether direct or by 44
means of telescience techniques. The committee believes that human-tending of experiments through telescience is likely to prove a productive and cost-effective approach over time. While many existing experiments, for example those currently manifested on the Spacelab, would be difficult to convert to make use of telescience at this stage, the committee believes that experiments planned for the 1992-1997 period should be designed to make effective use of telescience, where appropriate. It should be noted that the degree to which telescience techniques and apparatus will have to be tailored to individual experiments and not used in a multipurpose fashion is still somewhat of an unknown. The incorporation of telescience into the design of microgravity experiments likely will occur in an evolutionary manner. Presently, roughly 24 to 48 months are needed to adapt well-understood experiments so that they can be conducted in an automated fashion. However, because there is a poor understanding of many of the scientific processes involved in microgravity research, increased knowledge will be needed before teleoperated microgravity experiments become the norm and the majority of experiments can be carried out on a free-flyer. It should also be noted that the microgravity research culture will have to adjust to a new way of doing things if telescience is to become widely adopted by that community. In summary, current A&R/telescience technology can provide any information to a ground-based human observer that a video camera can see; it also can give the observer the ability to activate switches and valves on the space vehicle, reprogram its computers, and perform simple manipulations on the experiment using multiaxis remote manipulators. Eventually, computer-graphic displays with pull-down menus and active cursors may enable the remote human operator to elicit advice from the computer, get unsolicited warnings or other information in an understandable form, and make a variety of reconfigurations in an experiment. Given time and adequate resources, most microgravity experiments that can be completely rehearsed can be automated. Clearly, full automation and telescience techniques are essential if experiments are to be performed in a vehicle such as a CDSF where humans will not be present when many experiments are performed. 45
