National Academies Press: OpenBook

Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium (1987)

Chapter: Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems

« Previous: Teleoperation, Telepresence, and Telorobotics: Research Needs for Space
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 292
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 293
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 294
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 295
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 296
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 297
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 298
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 299
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 300
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 301
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 302
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 303
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 304
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 305
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 306
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 307
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 308
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 309
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 310
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 311
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 312
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 313
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 314
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 315
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 316
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 317
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 318
Suggested Citation:"Telerobotics for the Evolving Space Station: Research Needs and Outstanding Problems." National Research Council. 1987. Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium. Washington, DC: The National Academies Press. doi: 10.17226/792.
×
Page 319

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.

TELEROEO1TCS FOR THE EVOLVING SPACE STATION: RESEARCH afire AND CUTSIANDING PROBLEMS Lawrence Stark INTRODUCTION m e definition of robotics (TR) Hal not yet stabilized nor made the standard English language dictionary. I tend to use telercbotics as g remote control of robots by a human operator using supervisory and some direct control. Emus, this is an important area for the NASA evolving space station. ~ ~ ~ ~ ~ . ~ . e_. _ 8 ~ . ~ By robot, I mean a manipulator/mbbility device with visual or other senses. ~ do not name manipulators, as in many industrial automation set-ups, robots even if they can be flexibly p~ug~uu~35; rather calling these programmable manipulators. Our own laboratory at the University of California, Berkeley, has been involved In problems in display of information to he human operator, In crciblems of con~l of rate manipulator by he human Operator, and ~ rucat~on copays and bar~id~ch limitations as influencing both control and the display. A number of recent reviews have appeared with discussions of the history of Robotics beginning with nuclei' plants and underseas oil rigs. THREE SIMULTANEOUS RESEARCH DIRECllONS , __ ~ _ _ _ , There are problems using man alone. I believe that we should engage in triplicate or three way planning. It is important to carry out our research to accomplish forks (i) with man a Lone, if possible, such as in EVA extravehicular activities), (ii) with autonc~us robots (AR), and (iii) with t=]erdbotics. By comparing arxt contrasting the resort n~-=sary to carry out them three approaches, we may clarify our present problems. (S~ Table 1) the space environment is hazardous. It is very expensive to have a man In space; NOVA must have quite adeq late cost figures obta Bed from the demonstration projects that have already been acccmplished with the shuttle program. We may also need a higher quality of performance than man alone can provide in terms of strength, resistance to fatigue, vigilance, and in meeting special problems. For example, if the space suit is not of constant volume under flexible changes of the limbs, then a great d=~1 of strength is used up just in maintaining posture. 292

293 CABIN ~ Triplicate Plan m ng Problems with man alone Hazar~cus environment: (space similar to nuclear plants, underseas) Expensive (i.e. Eve ~ space) Need ~ncr=asP~ quality in Strength Fatigue resistance Vigilance Performance Problems with Autonomous Robots Not yet available Design not fixed Feasibility not certain Reliability not tested Therefore: TR is a viable leading edge technology All three directions should be sup ported for evolving space station planning, research, and development. Problems with autonomous robots lie in cur not having mastered the technology to build them and have them perform satisfactorily. They are not yet available! Indeed, designs are not yet fly arm it is not chain how feasible they will be, especially In terms of rdbustr~ess arm reliability. Therefore, we can ~ that t~lerabotics is a Friable leading edge technology. However, all three dir ~ ions should be intensively pursued in research and development, especially for the next stages of the evolving space station plane mg. SPACE SIAIICN TASKS One of the major roles that NASA can play is to hypothesize tasks for the evolving space station. In this way research regarding the design of t=]erabots to accomplish these tasks can be guided. For a list of seven groups of tasks see Table 2. \

294 TABLE 2 NOVA should Hypothesize TASKS for Evolving Space Station Hcusekeeping Life support systems Inventory control, access and storage Record keeping Garbage disposal Protection From space garbage Fin meteorites From traffic flow maintenance Satellite Vehicles Space station itself Construction Additional space station structures Manufacturing Crystal growth, biopharmaceuti~als Mobility Automatic piloting Navigation Path planning Scientific Lan~sat type image processing for agriculture Meteorology Astronomy Human factors research Scientific record keeping As I will consider' lakers it is important to distinguish between those tasks unique to the NASA/evolving Space Station and those with "industrial drivers" that will accomplish development of near technologies In hopefully a superior fashion and thtlC ermble consecration of limited NOVA resources.

295 EMBLEMS IN TELEE;~:)BOllCE; First I overview problems in telerdbotics: those concerning displays, vision arid other senses (Table 3) and those dealing witch control art rmn~nication (Table 4). In each tahie, ~ start with basic properti== of the human operator and end lo? with planned capabilities of autonctnom; robots. In between, I try to Cover what knowledge exists new in our field of telerobotics. ~ rimental Set-Up for m rce-Axis Pick-and Place Tasks A teleoperation simulator constructed with a display, joysticks, and a computer enable] three-axis pick-and-place tasks to be performed and various display and control conditions evaluated (Figure 1~. A vector display system (Hewlett-Packard 1345A) was used for fast vector drawing and updating with high resolution. An our experiments, displacement joysticks were mainly used, although in one experiment a force joystick was used to compare with a displacement joystick. An LSI-11/23 computer with the RT-ll cgerating system compute was connected to the joystick outputs through 12-bit A/D converters, and to the vector display system through a 16-bit parallel I/O port. CABIN 3 Display Problems for the Human Operator Display graphics (raster/vector) Cn-the-screen enhancements On-the-soene enhancements Other senses displayed Inputs to other senses Perspective and Stereo Displays Task performance criteria Helmet Mounts Display Telepresence; space constancy Human Operator (H.O.) Performance Fatigue, effort, vigilance Robotic Vision LLV - mips MLV - blockworld and hidden lines HLV - ICM, AI

296 ]~E 4 Control and C~nication Problems for the Mean Operator Chic properties of H.O., Medially for EVA task performance ITerve, Muscle, AG/A~ model Sampled data (SD) and adaptive control Eviction, preview, optimal control--~lman filter H.O. control of vehicles, mamal control He 0. control of TR H.O. Special control: Previewer, delay, bilateral, h~nor~ihiccontro Ideation (human, robotic): Navigation ~ pathways Potential field algorithms HIC (high level control): Supervisory control MUltiperson cooperative control; RCCL; fuzzy sets Autonomous robotic (AR) control Sensory feedback, adaptive control, AI A typical presentation on the disolav screen for three-axis . ~ . . p~c~-an~-p~ace tasks ~nc~udeu a cvlin~ri~1 manipulator. objects to _, ~ ~ ~ _ . ~ . _~ ~ ~ _ ~ ~ ~ ~ - - - J ~ plCK UPJ and Boxes In w m an to place them, all displayed in perspective (Figure 2). . . . ounce perspecclve pro~ec~lon atone is not sufficient to present ~nree-dimen~sionA' information on the twc-dimensional screen, a grid representing a horizontal base plane and references lines indicating vertical separations from the base mane are also Presented {~1 l;c ~! =1 BOW. I; ~. =1 `~= =~ ~ - ~ . / do_, ~~` = - mu., 1985 sUl~nitt~) . ~ ~ ~ . . The human operator controlled the manipulator on the display us mg two joysticks to pick up each object with the manipulator gripper and place it in the corresponding box. One hand, using two axes of one joystick, controls the gripper position for the two axes parallel to the horizontal base plane (grid). The other hand, using one axis of the other joystick, controls the gripper position for the third axis Overtires height) E~r~icu~ar to ache base plane. Picking up an Object is accomplished by touching an abject with the manipulator gripper. Liaise, placing an Object is a~nplisl~ed by touching the correct box with the manipulator gripper.

297 HP 1345A Cal I Graphic Display 4K x 16 bit Memory _5Z~ l \7 Joysticks J FIGURE 1 F~t=1 It. Patina Arm Simulator LSI - 1 1/23 Computer Pare bed I/0 12-bit a/D Converter I In addition ~ the gylir~ri~al manipula=r simulation, the kinetics are Tics of a six ~f-fre"am Rma robot awn were si~a~. Each of ~ ese degrees of fre - An were controlled silrmltanecusly using too joysticks;. Alff~algh no experiments have yet been performed with the puma s;~1ation, it is hoped that it will be a step toward experiments inch more complex manip~a~rs. A l~n~idth tan ep hone connection to control two Puma arms at Jet Propulsion Labs in Pasadena is plan net. The s;~1ation will allow prediction of the robots' Notion to provide a preview display to help overcome the communication delays inherent in such a low bandwidth connection, or as In transmissions to manipulators in space. .

298 l - T n 1 I FIGURE: 2 P~s-Berkeley visual errant display. Helmet M=nted Display Design Motivation The motivation of the HMD system is to provide the human Orator with a Iced epresence fey ing that he is actually In the rate site art controls the telemanipulator clirec~y. The HO system deters the h ~ n ~ rator's head motion, end controls the r ~ te stereo ~ ra accordingly. In cur current system, the remote -~1emanipulation task environment is simulated and the pictures for the display are generated by the computer. Head Orientation Sensors A twc-axis magnetic Helmholtz coil arrangement was used as a head orientation sensing device, to detect horizontal and vertical head rotations (Figure 3). By assuming that the pan and tilt angles of a remote stereo camera are controlled in accordance with the horizontal and vertical head rotations, respectively, the cc mputer generates the corresponding stereo picture for the HMD. m e head orientation sensing

299 Also ~1 I' it_ At_ it' FIGURE 3 Head orientation sensor device. LACE \ ~ device IS composed of a search (sensing) coil mounted on or beneath the helmet and two pairs of field coils fixed with respect to the human operator's con Loll station. me right-left pair of the field coil generates the horizontal magnetic flux of a 50 XHz square wave. m e up-down pair of the field coil generates the vertical magnetic flux of a 75 XHz square wave. The search coil detects the induced magnetic flux, which is amplified and separated into 50 and 75 XHz components. m e magnify ~- of each frequency component depends upon the orientation of the search coil with respect to the corresponding field coil (Duffy, 1985~.

300 Ton Display An early configuration of the HMD had a flat-panel Urn (liquid crystal display) screen (a commercially available portable Ton television) mcunt^~ on the helmet for the display (Figure 4). However, the picture quality of the Ton screen was poor due not only to low resolution but also to poor contrast. CRY Display A new design of the HMD that we currently have, mounted a pair of Sony viewfinders (MbJel VF-208) on the helmet (Figure 5). Each viewfinder has a 1- inch CR~ (cathode ray tube) screen and a converging Tense through which the human operator views the OK screen. The cc mputPr-generat-~ stereo picture pair (stereogram) is displayed on the CAT screens; one for the left eye and the other for the right. m e converging lens forms the virtual image of the stereogram behind the actual display screen. When the OK screen is 4.2 cm apart from the lens whose focal length is 5 cm, the virtual image of the Cal screen is formed at 25 cm apart frees the lens with an image magnification of 6. Emus, a I-~nch OK screen appears ~ be a o-lrx:n screen co me viewer. At appropriate geometrical and Optical cor~itions, the right and left images overlay, and most people can fuse the two images into a single three divisional image. me ster~ic display formulas use to generate the ster ~ r ~ for the hen - t mounted display are described in references (Kim et al., 1987~. LIGHT SOURCE SUPPORT CHANNELS / / a.. FIGURE 4: Folly HMD design with Ton screen. LCD DISP ~ Y

301 ! \~: ,. ~ ? AL, FIGURE 5 Current He design. M~hani~1 Design Five degrees of France were provided for the Satanical adjus~nent of the position and orientation of each viewfinder, allowing three orthogonal slidings ark two rotations (Figure 5~. A 1 Ib. counterweight was attached to the back of the helmet for count~r-balancing. Communication Delay and Preview Communication delay is a significant constraint in human performance in controlling a remote manipulator. It has been shown (Sheridan et al, 1964, Sheridan, 1966; TcmizuXa and Whitney, 1976) that preview information can be used to improve performance. Stark et al. (1987)

302 demonstrated that pr ~ few can significantly reduce error in tracking experiments with Imposed delay. Experiments were performed to investigate whether a preview display could improve performance ~ pick-and-place tasks with delay. A single bright diamond-shaped cursor was added to the display to represent current joystick position. This was a perfect prediction of bat the ~ effecter position wand be after the delay Sternal. ~us, the task was the same as if there were no delay, except that the H.0. had to wait one delay period for confirmation that a target had been touched or correctly placed (in the non-pre~riewed display, the target letter was doubled when picked un. and hem finale In when Of ~ the correct box). _ . . 4. , ~ ~~ ~ I__ _~ =~ ~ preview improved performance at delays up to 4 seconds so that it was almost as good as for a small delay of 0.2 seconds (Figure 6~. While tack completion time ~ the delayed condition increased greatly with delay, them was only a small increase in the preview case. This is because the H.O. Toast Sate for delays by using a '~ve-and~ait" strategy, making a joystick movement and waiting to s=- the resultant and effecter movement. In the preview case, this ax U; o _' o . - U it_ CL o 45Q 400 35D 300 Z50 200 150 100 50 / / / J / J ~ _ 0.O O.5 1,0 1.S TO 2.5 3.O 3 5 4.D De ~ By T i me (seconds) FIGURE 6 Performance affected by delays and by preview control mode.

303 scrap is only necessary when very close to the target or box to wait for confirmation that the goal has indeed bed toubbed. Coned Mae Experiments Position ark rate controls are the do can nap control modes for controlling tel~nipulato:rs with joysticks (or hard controllers) (Jensen and Corliss, 1971; Heer. 1973~. on the position control the joystick Inane ironical the desired era effe~or position of the manipulator, whereas in the rate control the iovstick Marx indicates the desired em effecter velocity. · — , -" In our three-axis pick~and-place tasks, the human operator controls Me manipulator hand position in the robot base Cartesian coordinate by using three axes of the two displacement joystick;. In pure (or ideal) position control, the system transfer function from the joystick displacement input to the actual manipulator harid position output is a constant gain ~, for each axis. In pure rate control, the syste`` transfer functiEn is a single integrator Is for each axis. In the rate control, a 5% dead-band nom Wearily Is introduced before the pure integrator in order to inhibit the drift problem associated with the pure integrator. Comparison of Pure Position and Rate Controls Three-axis pick-and-place tasks were performed with both pure position ark rate control modes for various gains (Figure 7). The Ian completion time plot clearly shows that pick-and-place performance with pure position control (mean completion time 2.8 seconds at Gn=2) was about 1.5 times faster than that of the pure rate control (~an completion time 4.3 seconds at Gv=4). Trajectories of Joystick and Manipulator Movements An order to exam me why the position control performed better than the rate control, several trajectories of the joystick displacement input and the manipulator hand position carpet during the pick-an5-place operation were observed. Typing trajectories flus the start of trying to pick up an object to its accomplishment were platted to illustrate position, rate, and acceleration controls (Figure 8~. Components only for the x-axis (side-to-side) are plotted, since components for the other two axes are similar. Observation of several trajectories indicates that a precise re-positio ~ of the manipulator hand is achieve] by a combination of quick step re-positionlng operations and slow smooth movement operations. En position control one quick step re-positioning of the manipulator hand from one position to another yes one joystick pull or push cFeration, whereas in the rate control it requires a pair of operations; pN11-and-push or push-and-pull operations (Figure 8~. This is a major reason why the

304 8 6 _ c' a, u, _' o 4 cot to cat at: RATE 't ~ ~ I; POS ITION ol I I I I I I I I I I 1 2 3 4 5 6 7 8 9 10 GAIN FIGURE 7 Performance comparison of position and rate control. position control yielded better performance than the rate control for Our pic~k-arxI-place tasks. It should be no, however, that the pic3~-and-place tank is a positioning task. If the task Is folla~ring a target with a constant velocity, then velocity (rate) control aced perform better. Acceleration Control Three-axis pick~and-place tasks were also tried with acceleration control. It turned out, however, acceleration control was not adequate to perform Sophie, safe pick-and-place operations. In acceleration control, the manipulator tends to move almost all the time even though the joystick is at the center position. Note that in pure rate

305 - J ~ INPUT OUTPUT POS ITION CONTROL J I 1 ~ . 1 1 , / ~ sec. / OUTPUT INPUT ~ ,*_~_% 1 1 t 1 1 1 1 a__ OUTPUT =~` INPUT RATE CONTROL - ~ I 1 1 1 1 1 ~ ! - ACCELERATION CONTROL EMIGRE 8 Position, rate and acceleration control. ~ ~__` I ~ 1 1 1 1 t I 1 1 J '. control, the r~nipula~r does not move when he joystick is at the center position regardless of prebills history of the joystick displacers. Oman Adaptation to Gam Mange Mean ca - letion time did not Garde math for the variants gains tested (Figure 7), With beans that the hewn curator adapted well to the gain He Amber et al., 1965; Young, 1969; S - ok, 19681. Both Icier and higher gains relative to the ppti~1 gains caused slight Increase In the nean completion time. A reason of slightly larger mean completion times with rawer gains is House rawer gains demand wider Justin displacements and it tam; lordlier for tile firmer or hand to display the joystick wider. A reason for slightly lordlier mean completion times with higher gains is that higher gains demand more minute joystick dippla~nts, degrading effective resolution of the

306 joystick control. An additional major reason for longer mean completion tin with lower gains for the rate control is due to the velocity limit. Force Joystick The two common joystick types are the displacement and force joysticks. The output of the displacement joystick is proportional to the joystick displacement, whereas the output of the force joystick (isometric or stiff joystick) is proportional to the fo ~ e applied by He human curator. me advantage of the force jc~ysti~ is that it requires only Ilunute joystick displacements (a few Caters) in contrast with the displacement joystick (a few centimeters). Pick-and-place tasks were performed for pure position and rate controls with displacement and force joysticks. The experimental results for two subjects (Figure 9) shows that in the rate control, task~performance with force joystick was significantly faster than that with displacement joystick. m is is mainly because the force joystick senses the applied force directly, requiring only very minute joystick displacements. In the position control, however, the force joystick performed no better than the displacement joystick. On fact, all three subjects preferred to use the displacement joystick in this mode, since the force joystick required more force to be applied than the displacement joystick, especially when the manipulator hand is to be positioned far away from the initial center position. Position control also performed better than the rate control regardless of joystick types, and furthermore the position control with the displacement joystick perforrr~l best for our pi~and-place tasks (Figure 9~. Resolution the experiments resets dernons~crate the superiority of position condom Hen He t=1emanipulator has a sufficiently small work apace (Figures 7, 8, & 9~. Note that cur ff~ree-axis pick-ar~place tasks t~ in this experiment implicitly assess ~t ache manipulator work apace is snail or at l=~t noLvery large, since our task allows the hen Tutor to perform successful pic3~-and-place operations with a display firm the entire work space on He semen. Examples of small work apace t=lemanipula~rs can be fed in nuclear reactor tPl~perators, surgical ~C~tpler~ts' or small deputes teler~tic hands. Position control can also be utilized during proxin,;ty cations in conjunction with the force-reflecting joysticks for Ranch t~lepr~nce (Bejczy, 1980~. Hen the telemanipulator's work apace is very large as ~r~ to Han pperator's con~cro1 space, position con ~ 1 of Me entire work space suffers from poor resolution since human operator's control space nest be greatly up-scaled to acoccmodate the t~lemanipulator's large work space (Flatau, 1973~. One way of solving this poor resolution problem in position control is using indexing (Johnson and Corliss, 1971; Argonne National Tech,

307 c) 4 I i_ ~ 3 As o o it 1 _ o .__ RATE it, - —`~, _~ POS ITION I I DISPLACEMENT FORCE JOYSTICK TYPE E]GURE 9 Displacement and force joystick control. 1967). In the indexed position control ~, the control stink gain is speck so that the full displa ~ t range of the control stick can caterer only a small portian of the manipulator work space, and large movements of the manipulator hark can be Inane by successive uses of an indexing trigger punted ~ the control stick. Nate, hover, that rate control can ir~tly praise any higher degree of res;ol~ion by Nate change of control stick gal shout ATOP of indexing. Ha~norphic Controller Mbst of our pi~k-arxI-place and tracking experiments were performed with juries as the input device through which the human Operator

308 controlled the simulated manipulator. The operator's mcvements when busing joysticks ~ e non-hcmec morphic, so that the mcvements he must make to produce a desired manipulator response do not match the movement of the manipulator end effecter. This, he must mentally convert the desired end effector position to Cartesian coordinates and use the joysticks to input these coordinate. To attempt to study whether a truly homomorphic input device could improve performance in tracking tasks, an apparatus of identical form to our simulated cylindrical manipulator was built. A vertical rod was supported by bearings on the base to allow rctation, theta. A counterweighted horizontal arm was attached to the rod with sliding bearings to permit rotation and translation in the r and z axes respectively. The human operator could control position through a handle on the end of the arm corresponding to the end effector of the simulated manipulator. Potentiometers measured movement in each axis to deters me input r, theta, and z. The LSI-11/23 computer read these values through A/D channels and displayed the manipulator in the identical position. Three-dim£nsional tracking experiments were performed with the hcmeomorphic controller and with joysticks for mains varying from 1 to 5 to compare performance (Figure 10~. . . . . . . . _ _ _ , _, _ , The results do not show a significant difference between the ncmeomorphic controller and joysticks over the range-of gain values. Although the larger movements required for the hcmeomorphic controller', with greater inertia and friction than the joystick, may have limited performance, we believe that human adaptability minimizes its advantages. Training by Optimal Control Example , A simplified simulation of the manned maneuvering unit, AMP, enable] study of training of human control performance (Jordan, 1985~. Only three translatory degrees-of-freedom, x, y and z, were used. Thrusters generating pulp== of acceleratory control were controller via a keyboard and the task was to accelerate simultaneously in x, y and z to a maximum velocity, transit to the desired new location, and decelerate again simultaneously. Two displays were used--a perspective display of a magnified model of the ~1, or two tw~i~nsior~ projectors of that model with a small inset of the perspective display. Subjects generally performed poorly Doris the few hurx~red seconds allowed for the tasks (Figure lla). It was decided to allow the subjects to view this control problem caries out by a simple optimal This ex~rierlm~ was of control algorithm (s ~ Figure lab). considerable help and several subjects then performed quite well t Figure llc). This experiment, learn mg-by-example, illustrates a strategy that perhaps may be effective in more complex and realistic funks as well.

309 i- ~ O. 8 O.6 L AL U. G. 4 O.2 O.0 FIGURE 10 Han~rEihic Con~croller. ,' l : _ /~ - '-E~~ Add. - ~ _~ ~ ~ t J 0 1 2 3 4 5 6 Ver-t i cc~ ~ Ga i n INIXJSTRLAL MOVERS FOR CERTAIN NECESSARY SPACE STATION TECHNOLOGIES e ~ . e _ . e _ _ _ This next section deals with the future' and e ~ cially with preindustrial driver sll other than NINA for new technologies which may be required in the evolving Space Station. In Table 5 I list nine components of a Robotics system that certainly seem to be driven by Important industrial hardware requirements, research and development. Therefore, it seems reasonable for NASA to sit back and wait for and evaluate these develcpments, saving its resources for those necessary technologies that will not be so driven. Looking at these figures gives ~~= some concept of how industrial develop may provide various types of technologies for the evolving Space Station; indeed' NASA may be able to pick and choose from off-the-shelf items! For example, the most powerful computers on the last space shuttles were the han5-held portable computers that the

310 o o o o . . . o ~ ¢ 0 ~ ~2 , o C\2 - Z ¢11~P o o o o o o ~ CU ~ CU ° ~ 1 UI - ~ ~I~P - I f o o o ~ ~ ~2 o o o o 0 CO ~ o ~ C\2 ~ 1 ~ Ul - X `-lI~P o .. _ tt, , _ . N o o o ~ _ 0 r' C\2 ~ _ ~ O ~ 1 O i ~D O - C" ·— o O C`2 _ ~ o o V - n o 1 o E o o . ~o/no _ ~ o~od Z 1 1 1. ~ 0 1 7jo/uo _ 9~1Od A ! ~ ; - 1 i ! ~ o , `'o/uo ~ ~3od X o - o C~ o 0 1 CD o CP~ o o o C\2 - e o 0 . o 1 o E CO I, o _ o - o ~2 - e o ~o o o 6 ~ ._ l~ ~\ £ - v -,' 0 o o ~_ -= ~ IL\ ~ £ ·. H

311 - ~ a o o o o N — ~ _ ~ ~ — O O O o o O J1O/UQ — ~0 J Z UI _ Z calf ~/~ — A?IlDOI9A Z _ S 1 ~ Lo Lit I I No. ~ + O I ~ O O . ~ O a a O N , N ~ ~ ~ id ~ ~ ~ ~ ]]O/UO _ 2 ~03 ~ fig ~ ~ ~ ~ rim Ul — ~ £'l1Dp ~~r ~ ~ it._ ~ + O , O LIZ CO [~] CO ~ O O o o 0 ~ ~ O ] ~O/UO - a DJ03 V [U - X ~2p S/LU _ ~°19^

312 o o o o o o o ~ ~ ~ ~ o ~ O2 ~ CU 1 1 UI - Z ~I9P 1 / 1 - - J - 1 ~ O O O ~ ~ C`2 O o o O 0 ~ O 1 1 ,~ _ ~ ~I~P / o o o o . . . o C~ , UI - X ~P o o · . ~ o C\2 ~ ~ 1 o o / - ~ . . _ ~ _ / ~ O ~ I I ~ ~ o . ,~o/no _ ~ ~od Z o - o C~ _ ~ o o 0 1 o C': ·— o o - o C\2 ~ — C o o o 1 o ~ C'7 ·— o _~ o C\2 _ o o C~ o 1 o ~ ._ 1 l tOD - ~ o 1 }~°/~° - ao~o~ 1 ' o - o CU - e o 0 CD 0 ~ ~ ._ o - . 0 ~2 _ o 0 CD o 1 ~D - 0 ~ C'~ ·- ],O/UO - BD103 X - C) _ _- _ _ ~ 1" 811 - (~ C~ £ - - - - ~' ~ .~ 11u ,5 at ~ ~d ~: 5

313 TABLE 5 Drivers other than MESA for Nine Ne~ed Technologies Robotic Manipulator arm Control S6h~ Joystick - Aircraft AR ~nufacturir~ Ir~ustry, Nuclear Irx~ustry, Minim Irxlustry, Sensors: Force and Touch; compliant control ROV and Mobility Military, tanks arm other vehicle plans? Undersea ROV - Oil are] C ~ nications ~ stay Lcoo motion - University Research Shipping Industry: Ships at Sea EAR, TR, M~n] TV Camera ~ , Entertainment Industry - commercial device Security Industry Need mounts, controls and motors for PAN, TILT and for Stereo VG Graphics Entertainment industry is a better driver than companies building Flight Simulators; HMD as an example. EM sensors research/Head-Eye ME use ICM Landsat Security Medical Industry - CT and MRI Industrial Production Lines TD - Image Understanding Computer Computer Industry (HDW) and (SFW) Computer Science research base i Communication now very broad Communication Industry is huge Ships at Sea BE Compression Remote Oil Rigs Arctic Stations Plans and Protocols to Combat H.O. Fatigue and to Promote H.O. Vigilance Office Automation Forces Air Traffic Control NOF5.C Security Industry Cooperative Control Military - submarme control Helicopter flight control Air traffic controllers Nuclear industry Chemical plant industry

314 astronauts bought aboard With contained huh greater capability than the on-board computers;; those had been frozen in their design ten year; ago in the planning stages for the space shuttle. NECESSARY TEIEEK)I~lIC~ IECHNOLOGTF~ 10 BE SPARED BY MESA However, there are several areas in telerobotics that may likely not be driven independently of NASA, or where MESA may have an important role to play. Indeed, the Congress has specifically mandated that 10% of the Space Station budget should be used for Automation and Rcbotics development, and that this in some sense should spearhead industrial robotics In the United States (Table 6~. TABLE 6 Areas Sparked by ROSA not industrially Driven Visual Enhancements for Graphic Display Telepresence with Stereo Helmet Mounded Display (HMD) Multisensory Input Ports: Worry about H.O. overload condition (especially with cooperative control and communication) Higher Level Robotic Vision: Example Image Compression by Modeling (IC~) (to require less information flaw and faster update) Special Control Modes for H.O. Hcmeomorphic control Bilateral control Time delay and preview control for time delay Compliant control Higher Level Control Languages (such as RCCL; fuzzy control; path planning by potential field construction) Remote operating vehicles (ROY) special control problems: Navigation, orientation, obstacle avoidance for ROV Cooperative Control: Cocperation amongst humans, telercbots, and autonomous robots Ccmpliant, Flexible, Hcmeomorphic Manipulators Grasp versus tool using Homeomorphic Dual Mode Control Impedance Control

315 UNIV~S11Y NINA RESEARC]I I now would like to make a plea that NASA should ex pen] and stimulate teleroboti~= research conducted within the university environment. Of course, as a professor I may have a bias On this direction and I am willing to listen to contrary arguments! In addition to the benefits of the research accomplished by universities, NEST also gets the education and training of new engineering manpower specifically directed towards telerobotics, and focused on the evolving Space Station. What kind of university and educational research should be funded in general by NASA. I believe there are two levels of cost (with however three directions) into which these educational research labs should be classified. (i] First are Simulation Telerobotics Laboratories. Here we need graphics computers, perhaps ~cyst~cKs, perhaps nigher level supervisory control languages, cameras, image compression techniques and communication schemes. I would guess that cur country needs at least thirty such systems for education and~rainir~. mese systems should be very ~ne~ensive, approximately $50,000 each. hey need not even be paid for by No=, since universities can provide SU*1 research simulation laboratories out of their educational budgets or fray small indivi~1 Church grants. Our Teler~ibotics Unit at Berkeley has been thus funded. A good deal of exploratory rerun can be carried Ant inexpensively in this manner. ( ii ) Secorxt, we need Telerobotic ~boratoriPs with whys ical manipulators pent as important r~r~ components. In this way, experiments with various rcibotic manipulators, especially those with Special control characteristics such as flexibility, h~norphic form, new devel ~ nts In gras ~ rs, and variable i ~ nce control m ~es, other than are found in standard industrial manipulators, would be possible. ~ guess that there are about five such laboratories On some stage of development at major universities in the country. ~ would further estimate that these laboratories could each use an initial development budget of $300,000 to enable them to purchase necessary hardware in ablution to software as existent in the Simulated Teleroboti~= Laboratories. Another set of costly laboratories would be Telerobotics Laboratories with remote operatoring vehicles (ROV). Here again, we need about five laboratories at universities with first class engineering schools. Again, ~ estimate about $300,000 each for the initial hardware support of these REV labs. They could then study transfer vehicles, local Space Station vehicles, R bon/Mars Ravers, and even compare MMU vs. Robotic controlled vehicles. m e university laboratories WoN1d contrast with and serve a different function than ongoing aerospace industrial laboratories, and MESA and other government laboratories. mese latter assemble hardware for demonstration and feasibility sues. men unfortunately they are somehow unable to carry out careful human factors research dealing with the changing design of such piers= of equipment. In the university setting, this apparatus could be taken apart, changed,

316 revitalized, modified arm the flexibility Ed inform our current capably i=. I wand like to contrast the Gossamer Condor and Gossamer Albatross with the NASA program. It was char that if Ready was ever to be sulfur, he had ~ build an experimental plane Rich was ~ to break down each experiments day. But He plane could be repaired few mimers! This "Laboratory beech" concept is so different from t~renty-y-~r-ahead-plannir~ currently controlling our space page ~ that ha= been effectively eliminated at NASA. I think it is important to reintroduce rough and ready field laboratories back into the space program. NMA MAZE; Another role that NOVA might play IS to offer demonstration contracts or, even better, prizes for accomplishment of specific tacks. Again I turn to the Kremer Prize; here a private individual donated prize money to be awarded to the first to build a man-paw aircraft conformirx~ to cercain carefully laid Ant Specifications. Fornication Barbels for controlling rate vehicles and Rae manipulators are already set up. Thus we cc=d have prize contestants demDnstratir~ at differing locations on earth at one "g"; next demonstrations using elements capable of cperatlng in space, or even more stringently, of having that minimum mass capable of being lifted into spare; and then we might have true shuttle and space station demonstrations. DUAL ~~ ~ TR FOR THE SPACE STATION Finally, I would like to leave you with the thought that the list of t~be-spark~¢y-N~A prcblens ~ Table 6 contains many important int~ll~tua1 prcd31ems facing the area of tel~otics. Although these areas are being approached ~ our regears Unity at the present time, it may not be possible to foresee what novel kirks of dhall~es will face the evolving Apace Station In twenty years;. Even though I may not predict accurately, ~ Mainly hope I am Scheme in person to watch t=1er~tics playing a major role In Operating the Space Station. i~Y hone I ~ =~e in An to SPRY The telerdbotic, IR, sync em is a simulated distant Abbot with vision and manipulator an4/or mbbilit-v subsystems controlled by a human cgerator, H.O. the H.O. is informed mainly by a visual display, but also by other sensors and other sensory displays, i.e. auditory, force or tactile. His control can be Direct via joysticks, or supervisory via command and control proves et-tec~ed by partially autonomous rc3:otic functions. Delays and bandwidth limitations In Fornication are key problems, complicating display and control (Stark et al., 1987) .

317 Class e~erim~ts Tabled our Telertibotic Unit at Me Un~versi~r of California, Berkeley to explore in a mmixr of ~ directions. The ~ direction has now been greatly exterxied and is a major focus In our laboratory. C'n the other harm, the homomorphic controller' did not seem to be a pr Inductive project to continue because of the adaptability of the H.O. to many configurations of control. Also, our interest in supervisory and other high level controls is beading us away from the direct manual control. me students taking a graduate control course, ME 210 "Biological Control Sybems: Telerobotics," during the fall semester, 1985, in which the helmet mounted display, HMD, is emphasized, were enthusiastic and felt the course stimulated the ~ creativity and provided an opportunity for them to engage in relatively unstructured laboratory work--a good model for subsequent thesis rem. ACKN~ We are pled to acknowledge support Frau the NASA-Ames Pe#earch Center (Cooperative Agreement NoC 2-86) and the Jet Propulsion Laboratory, California Institute of Technology (Contract #956873). We Could also like to thank visiting lecturers from NASA-Ames; park Cohen, Stephen Ellis, Scott Fisher, Arthur Grunewald, John Perrone and Mbrdeccai Velger; Drs. Won Soo Kim and Blake Hannaford, and Frank Tendick, Constance Ramos and Christopher Clark of University of California, Berkeley. . , ~— . ~—, ~ ~~ Argonne National Laboratory 1967 Manipulator System for Span Applications. T - hnical Report, A~orme. Bejczy, A. K. 1980 Sensors, controls, and man-machine interface for advance t~lecperation. Science 208~4450~:1327-1335. Duffy, M. K. 1985 A Head Monitor System Using the Search Coil Method. Master's thesis, Department of Electrical Engineering and Computer Sciences, University of California, Berkeley. Ellis, S. R., Tyler, M., Kin, W. S., M~Greevy, M. W., and Stark, L. 1985 Visual enhancements for perspective displays: perspective parameters. Pp. 815-818 in A::: Proceedings of the International Conference on Systems, Hen, and Cybernetics.

318 Flatau, C. R. 1973 Ibe manipulator as a nears; of ~ our d~s capab; ~ ities to larger arm smaller scales. Pp. 47-50 In P~ings of 21st Conference on Pate System Tedhnolo=. Ha', E. 1973 Perwtely Mewed Systems: Exploration and Operation In Space. ~1 ifornia Institute of T~logy. J~, E. G., and Corliss, W. R. 1971 Swan Factors Applications In leleaperator Design and Operation. Wiley-Intersci~e. Jordan, T. 1985 The Simulated Anne Maneuvering Unit and Pursuit ~rixnts. Mastery thesis! Department of M~ani~1 Engineering, l~iveristy of C~lifornia, Berkeley. Kim, W. S., Ellis, S. R., Tyler, M., Seafood, B., arm Stark, L. 1987 A quantitative evaluation of dative arm s~ic displays In his martial tracking tasks. l :. Trans. On System, Mi3n, arm Cybernetics 16:61-72. Kin, W. S., Ellis, S. R., Tyler, M., ark Stark, L. 1985 Visual ~a~ts for t~ler~tics. up. 807-811 ~ I:: Pigs of the Int. Conf. on System, Man, and Comics. Kiln, W. S., T~ic~k, F., and Stark, L. 1986 Visual er~nc~nts ~ pi~and-place tasks: human c~rator's controlling a simian pylir~ri~1 manipulator. Submitted to the i::: Journal of Robotics arx]Automation. Miner, D., Graham, D., Mel, E., ark P~eisener, W. 1965 than Pilot dynamics In C~nsatory Systems: Ohm—~ Models, and Experiments with C:on~lled Elements and Foxfire Fashion Variations. U. S. Air Force, A~D~R-65-15. Sheridan, T. B. 1966 Three newels of Paris' consul. ~ Hewn , _ Factors ~ Electronics H=-7: 91-102. Sheridan, T. B., M~1, M. H., et al. 1964 She predictive ads of the human controller. Progress in Astronomics and Aeronautics. Vol. 13. Academic Press Inc., N - r York. Stark, L. 968 Neurological Con~cro1 Systems : Pit Em. Studies in Bioer~in~rir~.

319 Stark, L., Kin, W. S., lyrics, F., et al. 1987 Teler~tics: display, control and fiction problems l ~:~:~: Journal of Robotics and Automation RA-3 (1) :67-75. . Tcmiz~a, M., arm ~itr~ey, D. E. 1976 me human Operator ~ previewer traduce: an ~pe:riment arc its maleling via septic control. Transactions In the ASME Journal of Dynamic Systems, Measurements, arc Control 98:407-413. Yours, L. R. 1969 On adaptive mat control. Systems M~-10 (4): 292-331. lo::: ~=iO~ ~ ache

Next: Discussion: Comments on Telepresence and Supervisory Control »
Human Factors in Automated and Robotic Space Systems: Proceedings of a Symposium Get This Book
×
Buy Paperback | $125.00
MyNAP members save 10% online.
Login or Register to save!
  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!