“Exploration Telepresence: Value and Challenges”
Total Page:16
File Type:pdf, Size:1020Kb
“Exploration Telepresence: Value and Challenges” Dan Lester University of Texas FISO Telecon May 27, 2015 Human cognition makes use of “presence”! in distant places. Cell phone, Skype, etc.! Once achievable only with human bodies. ! Congress endorses human “presence” in spaceflight.! ! !!!!Telepresence! The cognitive mandate for telepresence! depends on ! • control latency ! • control bandwidth! Telepresence and human space flight! as enablers for space exploration.! Human-robot collaboration! FISO Telecon 5/27/15 2 The idea of space telepresence was considered in some detail at the Exploration Telerobotics Symposium, May 2-3, 2012 at GSFC. This symposium explored Space Exploration via Telepresence: A New Paradigm for Human-Robotic Cooperation See presentations and final report at http://telerobotics.gsfc.nasa.gov FISO Telecon 5/27/15 3 Recipe for Telepresence! Control bandwidth for quality telepresence?! • Human vision ~10 Mb/s per eye! • A bit more with haptics, dexterity, and audio! (with compression, and some generosity, let"s say 1 Mb/s)! Control latency for quality telepresence?! • Threshold for latency detection is ~50 ms but …! • Human reaction time for visual stimulus is ~200 ms! That defines a “cognitive horizon” of c * 200 ms / 2 ~ 30,000 km! But everywhere we want to go is way beyond 30,000 km from Earth!! • Earth-Moon 2.6 seconds • Earth-Mars 8-40 minutes! WE CAN"T HAVE “QUALITY TELEPRESENCE” AT THESE ! LOCATIONS FROM THE EARTH. BUT WE CAN IF WE GET CLOSE.! FISO Telecon 5/27/15 4 But let"s talk about latency! Dave Scott’s Apollo 15 stopwatch FISO Telecon 5/27/15 5 What"s wrong with a little control latency, anyway?! For driving around, and avoiding obstacles, not much. ! (10-20s latency endured by Lunokhod controllers made driving around hard, but they did it) ! Few careful assessments of the latency penalty on cognition for >1 second.! Oscillopscia (perceptual instability relative to inertial and gravitational frame of reference)! and dynamic head-tracking errors. Latency screws with your brain. ! Allison et al. (2001, 2004) assessed ! 22.5 deg/s perceived stability by a user with a stereo! head-mounted display while panning head.! 45 deg/s “Additional tracking delay” (abcissa) is the! added latency between the motion of their! head, and the motion of the image.! 100ms two-way latency isn"t that bad.! 90 deg/s Just go slow. FISO Telecon 5/27/15 6 What’s wrong with a little control latency, if we’re trying to do something? Well, it depends on what you’re trying to do. If you"re just trying to guide a 7 DOF arm to do a positioning task, it"s not that hard, but there are latency penalties.! Lane 2001 saw that the required completion time using an “unmitigated” situation, where the user simply saw what was going on, rises steeply with latency. ! It helps if you have predictive assistance, and also if you have a graphical indication of what you commanded. ! FISO Telecon 5/27/15 7 What’s wrong with a little control latency, if we’re trying to do something? It depends on how hard it is. Piece of telerobotic history! -- work of Bill Ferrell, back in the 1960s. ! Simple 2D “minimal manipulator”.! Task completion time versus latency and “difficulty”.! “Difficulty” defined as how far you had ! to move, divided by how precisely you got there. ! Not surprisingly task completion time depends ! critically on latency and difficulty. ! FISO Telecon 5/27/15 8 What"s wrong with a little control latency, if we"re trying to do something? How about doing really hard stuff?! Telerobotic surgery is exquisitely difficult. High-dexterity w/ precision cutting and stitching! of compliant material. Low error tolerance. ! A LOT of work assessing latency penalty. 500ms is considered practical limit.! Assessment by Rayman 2006 – ! task time for average of four! laproscopic surgery manipulations.! Things get bad fast beyond 500ms.! Errors also go up quickly.! Lesson of telerobotic surgery is! for high dexterity operations, 500ms ! is a practical limit for latency.! FISO Telecon 5/27/15 9 … and then there is easy stuff that is hard (unlike surgery) The NEEMO crew experimented with undersea habitat-to-shore telerobotics, and found that tying a shoelace with Earth-Moon latency (2.6 seconds) took almost ten minutes. FISO Telecon 5/27/15 10 What’s wrong with control latency, if we’re trying to do something? What “hard stuff” are we talking about? Picking up rocks is hard with latency.! Different sizes, random shapes. Turning over to inspect.! Assembly and maintenance is hard with latency. ! Drilling, digging, stacking and cabling.! In general, dexterity is hard with latency. Situation awareness as well. ! Haptic feedback extremely intolerant of latency. ! For field geology, we have NO experience with low latency telerobotics. ! Lessons from high latency telerobotics (MER, MSL) don"t ! necessarily transfer well to low latency telerobotics. ! Analog studies on the Earth will be essential to this understanding.! Latency makes life HARD. ! So let"s do something about it. ! (But landing humans in gravity wells is hard too.)! FISO Telecon 5/27/15 11 So what are we talking about? On-orbit telerobotics.! Human “presence” from a (short) distance away. LOW LATENCY.! (images from Boeing, GSFC, etc.)! FISO Telecon 5/27/15 12 Controlling telerobots from space microgravity! may pose some challenges …! Human factors are an important element of exploration telepresence. ! • Sensorimotor deficits in eye-hand coordination in microgravity, “space-fog”?! • Cognitive complication of control in a gravity field by an astronaut in microgravity.! • Space habitat fatigue and work overload on cognitive functionality.! • Teleoperation control station design in microgravity – display screens, dexterous manipulators.! • Candidate screening & skill assessment.! HARI – NASA ARC/JSC !Otmar Bock/DLR! Chuck Oman/MIT ! !PMDIS - Barry Fowler/York U. ! Doing telerobotics in space is serious issue for exploration telepresence.! ISS probably at least as important for assessing ! human factors as technical factors. ! (Canadarm/DEXTRE not that relevant anymore.)! FISO Telecon 5/27/15 13 6 Crew Human Exploration Using Real-time Robotic Operations (HERRO) Schmidt, Oleson, Landis et al. Glenn Research Center FISO Telecon 5/27/15 14 Low latency telerobotic control process?! Undersea oil, gas, cable Drones/UAVs Mining Lessons about telepresence! architecture from terrestrial ! applications.! Surgery FISO Telecon 5/27/15 15 Evolution of space ! telepresence! Lunar workstation at Lunokhod control from Earth cis-lunar habitat - latency ~few x 100 ms! (control latency ~5-10 seconds)! Space "##! telepresence ISS Remote ! enabled Workstation ! control latency ~ ms)! Boeing FISO Telecon 5/27/15 16 Low Latency Telerobotics - not a new concept for space exploration! Paine Report (1986) “Pioneering the Space Frontier”! "We recommend that: NASA explore the limits of expert systems, and tele-presence or tele-science for remote operations! including ties to spacecraft and ground laboratories. In working toward these goals, a broad examination of the non-space ! applications of tele-science should be included.” ! “Telepresence: The use of real-time video communications coupled with remote control techniques which would provide ! an operator on Earth's surface or other location with the capability to carry out complex operations in space or on the ! surface of a planet or moon.”! Stafford Report (1991) “America at the Threshold: America's Space Exploration Initiative”! “ "Telepresence# robots can conduct some geologic field work. While such a technique might greatly enhance the scientific ! return, the details of how such robots might work with people remain to be developed. Operators for these telepresence ! robots need near-instant radio contact with the robots. This may be marginally obtainable by having the controllers on the ! Earth, but operators on or near the Moon have a near-zero time lag for robotic teleoperations.”! and more recently… (including HERRO!)! Mars DRM5 (2009)! “Low-Latency telerobotics operation may be a useful strategy, particularly if human missions stay out of gravity wells for some time.”! Augustine Commission (2009) “Review of United States Human Space Flight Plans Committee”! “The goal focuses human exploration on producing exciting new science at each step of the way. The emphasis would be ! on obtaining multi-kilogram samples from a variety of solar system bodies through tele-robotic exploration in concert with ! The human missions. In the case of the Moon and Mars, humans would remain in orbit.”! “A Phobos-based teleoperated exploration of the Martian surface, returning with samples from that surface, would likely! precede a crewed Mars landing mission, and would provide dramatically more responsive remote control than with the ! communication delays incurred between Mars and Earth.”! NASA Technology Roadmap TA04 (2012)! ”Top technical challenges in human-robot interfaces are full immersion telepresence with haptic, multi sensor feedback, understanding! and expressing intent between humans and robots, and supervised autonomy of dynamic/contact tasks across time delay.”! FISO Telecon 5/27/15 17 and most recently …! Human Spaceflight Architecture Team (HAT) DRM 8a (2013)! Mars Orbital Mission (crew tele-operates surface assets)! Global Exploration Roadmap (2013)! “New mission concepts, such as human-assisted sample return and tele-presence should be further explored, increasing understanding of the important role of humans in space for achieving common goals.”! “Human-assisted sample return