Click for next page ( 9


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 8
Three Generations of Robots The industrial robot, its image having been widely promulgated in the print media and on television, almost invariably comes to mind today at the mention of the word robot. It has largely displaced the bipedal tin men of early science h~c- tion. The typically squat, one-armed, occasionally mobile f~rst-generation robot orig- inated in the 1960s. Usually, it was single purpose and was used in such occupa- tions as welding, painting, and machining. Today such robots are in wide use, having matriculated through the early stages of laboratory development and tech- nical feasibility to economic feasibility in the early 1980s. A second generation of adaptive, sensor-based robots, at the laboratory stage in the 1970s, are just arriving at the stage of technical feasibility. These are diverse robots, with some intelligence, but still largely single function. Like first-genera- tion robots, they are used primarily in manufacturing. A third generation of robots is needed to work outside the factory. The indus- trial robots employed in manufacturing operate in highly structured environments. Often the manufacturing environment is altered to accommodate them. Altering to any great extent the environments that service robots will be called upon to operate in (e.g., undersea and constriction environments, space, mines, nuclear power plants, hospitals, offices, homes) is inconceivable. To function in such unstructured environments, robots must come to resemble humans more than machines, including possession of certain humanlike faculties. Communication functions will be essential in these third-generation "intelligent" robots. Tele-existence (often referred to as "telepresence" in the United States), a sub- set of third-generation robotics, aims at natural and efficient remote control of a 8

OCR for page 8
9 robot by providing the operator with a real-time sensation of presence. It is an advanced form of teleoperaiion that utilizes rich sensory feedback to enable a human operator to function as one with a robot working in an inhospitable (e.g., highly radioactive, extremely hot, or oxygenless) remote environment. The pilot- aircraft interface in some advanced military fighter jets is a form of tale-existence. A system for tale-existence would comprise intelligent mobile robots, their supervisory subsystems, and remote-presence and sensory augmentation subsys- tems that would enable the operator to derive a computer graphics-generated, pseudorealistic sensation of presence from the robots' ultrasonic, infrared, and other, otherwise invisible, sensory perceptions. Realistic visual, auditory, tactile, kinesthetic, and vibratory displays would be realized in the remote-presence sub- system. A human operator would be capable of both monitoring a robot's work ~ through its sensors and interceding as necessary to perform the task vicariously as if conducting the work personally (ideally) or from within the robot (practically). The principle of the display that provides a sensation of presence can perhaps most easily be understood in terms of vision. The system is based on the principle that the world we see is reconstructed by the brain using only two real-time images on the two retinas. The environment supplies to the retinas two-dimen- sional images that change in real time in relation to the movement of the eyes and head. We reconstruct the three-dimensional world in our brains and project it onto the real three-dimensional world. To realize a robotic display, we would have to precisely measure human movements, including those of the eyes and head, in real time; construct anthropomorphic sensors and effecters that mimic our own in function and size; ensure that they can be controlled to follow precisely the move- ments of a human operator; and assure that the operator sees the same visual space observed by the robot by delivering the images recorded by its sensors directly to the operator's eyes. The display must enable the operator to see the robot's upper extremities, which would be controlled to track in real time precisely the same movements as the operator. Tele-existence has been proposed in principle, and its design procedure has been explicitly defined by the Mechanical Engineering Laboratory of the Ministry of International Trade & Industry. The laboratory has built experimental visual display hardware and demonstrated the feasibility of a visual display with the sen- sation of presence in psychophysical experiments. The laboratory has also con- structed and evaluated a prototype mobile televehicle system, capable of being driven by means of auditory and visual sensation of presence. A visual tele-exis- tence simulator, incorporating a pseudo-real-time, binocular, solid model robot sensor, has been developed and its feasibility has been experimentally evaluated. The feasibility of the tale-existence master-slave system also has been demonstrat- ed. This system involves an anthropomorphic robot mechanism with an impedance-controlled active display mechanism and head-mounted display whose structural dimensions closely approximate those of humans. Economic feasibility for telepresence is perhaps 30 years in the future, but the need for it is current. Work in telepresence is being carried out in the United States by the National

OCR for page 8
10 Aeronautics and Space Administration (NASA) and the U.S. Air Force, which are jointly developing a glove device capable of measuring, storing, analyzing, and displaying individual joint angles for each finger of an operator's hand and trans- forming these data into commands that correspond to arbitrary mechanical hand configurations. "Glovemaster" is designed to enable an operator to directly drive an analog servo system in critical applications, such as those found in nuclear power plants and space exploration. The device is accurate to three decimals and is an order of magnitude more reliable than seemingly similar devices developed for the electronic games market. The crux of application opportunity, however, lies in autonomous robotics. Autonomy has been central to past successes and is key to future success in ever more demanding service roles. One prototype service robot in use in the United States, a successor to the 6-year-old robot developed to explore the nuclear acci- dent at Three Mile Island, is capable of 35 motions that enable it to saw, handle, and perform water jet cutting (useful in accidents and demolition) and walk and steer over all terrains. The teleoperated robot is equipped with onboard computing and artificial vision and is capable of self-diagnosis. Robots with long arms, robots with dual arms, and walking robots with long slides needed to operate on rugged terrain where wheels will not work are under development by NASA for space exploration. One such robot, being developed at Carnegie-Mellon University, can take steps several meters in length using legs that move through the body rather than around it. NASA is also working on semiau- tonomous robots capable of operating on their own for the 10 minutes it takes radio transmissions to travel between the Earth and Mars. Land waste in the United States, estimated to be at least a $100-billion problem, has made excavation an immediate problem for robotics. Excavation robots are simple, slow, and powerful and are feasible with current technology. Eventually, intelligent robots might replace manual labor in strenuous and boring product acquisition work in farming and mining; deep-sea products, for example, might be acquired by robots transported by submersible vehicles. Generally, outdoor applications of the robotics technology that was developed for manufacturing are limited by lack of geometry and color differentiation and the difficulty of preplanning tasks. Indoors, the problems associated with lack of structure are somewhat more amenable to less radical adaptations of advanced industrial robot technology. Toshiba, for example, is developing robotics systems for mail feeding, address block location, and irregular parcel singulation and imag- ing that overcome significant limitations of existing mail sorting automation, which requires that pieces be uniform in size, shape, and address location. The system incorporates feeder components, vision systems and range sensors, con- trollers, conveyors, and a robotic arm and vacuum gripper. For the U.S. Postal Service, which currently incurs 20 percent of its total costs in manual sorting and expects its current volume of 140 billion pieces per year to nearly double by the year 2000, the system holds great economic promise. Intelligent robots could replace sales personnel at point-of-sale terminals at

OCR for page 8
11 gasoline stations; fast food emporiums; and department, hardware, and grocery stores. Commonplace chores of commercial establishments, such as cleaning (of factories, warehouses, stores, hotels, and office buildings), delivery (of mail, sup- plies, and medicine), and security (surveillance and fire protection), could be done by personal robots. Such robots could be utilized in the home for many of the same chores as well as for cooking and retrieval. Most of the application functions needed by personal-use robots those that require the robot to verify, inspect, recognize, locate, grasp, transport, store and retrieve, navigate, and move (robot and end effector~have yet to be developed. Meeting the ubiquitous requirements for reliability, speed, and low cost will entail design for automation, standardization, environment structuring, contact and non- contact sensing, artificial intelligence, human-robot communication (by speech, display, etc.), collision-damage avoidance, error recovery, and end effecter design. Surgical robots are another possibility. Though much of the technology needed for surgical robots may be borrowed from manufacturing robots, greater integra- tion is needed. A surgical robot would be capable of analyzing a patient and simu- lating the procedure to be performed and of using sensing modalities to obtain feedback during an operation. Work on medical imaging systems is being done, and systems for engineering models of bones exist. Key problems in robotic surgery are man-machine interaction in the surgical situation, relating models to reality, verification, precision, operating room compatibility and sterility, safety, error recovery, and backup. The goal of recently completed work at IBM and the University of California at Davis, for example, is precise machining of bone in cementless hip replacement surgery. The current manual procedure for preparing the femoral socket (which uses a hammer and broach) leaves gaps of 1 to 4 mm between the implant and bone, and there is very little control over where the hole is placed. In contrast, dimensional accuracies in bone on the order of 0.05 mm have been demonstrated by a robot. In these applications, the surgeon uses computed tomography (C~ images to determine where the implant is to go and does the rest of the procedure normally. Such coupling of robotic precision with human intelligence to augment surgical procedures has broad potential application in such fields as neurosurgery (where robots have already been used clinically), plastic surgery, oncology, and ophthalmology, as well as in orthopedics. Safety and man-machine communica- tions are clearly vital factors in robotic surgery. Not only must there be many safe- ty checks, but the surgeon must remain in complete charge at all times. He needs to know what the robot is doing at all times and be able to intervene and possibly redirect the robot interactively. Robots with a "softer" touch, able to handle people rather than industrial machinery, could care for hospital patients and the elderly. The United States, for example, has a quadriplegic population of some 60,000, tragically augmented by 6~000 new individuals annually. Full-time care for these people, whose average postaccident life span is 45 years, is $100,000 per year. Personal robots could pro

OCR for page 8
12 vice such individuals with more consent and intent ate care at a lower cost than those for human attendants. The aging populations of both Japan and the United States could provide employment for vast numbers of such robots, relieving young people just starting their own families from the need to care for aging parents and aging parents from the mental burden of being cared for. Former U.S. Secretary of Health, Educations and Welfare Joseph Califano has estimated that for every month that entry of aged Americans into expensive and all too often dehumanizing nursing homes is delayed, there is a cost of $1 billion to the U.S. economy. Homes with wheelchair access will readily admit robots, and work on a "smart" house that will be more amenable to robots is under way in Japan. A feasibility study conducted in Japan revealed the potential of, and impedi- ments to, personal robots. Expectations were listed in He following order: house- hold chores, security, environment, health, recreation, and education. Actual applications were prioritized as follows: home, medical services, health, sports and recreation, education and culture, human relations, public service, and shop- ping. Conditions for popularity of the personal robot were economy (the cost of a robot must approximate that of an automobile), friendliness, social adaptability, and ease of maintenance. Technical hurdles include safety and reliability, flexibili- ty of operation, man-robot interface, walking or other mobility function, human safety, and energy supply. Although U.S. committee members articulated many of the same expectations of and applications for personal robots, actual development work in the United States seems more heavily oriented toward environmental (e.g., excavation, nucle- ar site, space exploration) and medical services (e.g., surgery) robots than toward personal robots. It may well be economics that determine whether third-genera- tion, intelligent robots capable of working with people in a symbiotic relationship will be used predominantly in health care and human services or in hazardous occupations, such as nuclear power plant work or undersea and space exploration.