2.2.4 Free Flying Servicers

This segment of the program is focussed on the development of space robotics for eventual application to on-orbit satellite servicing by free-flying servicing robots. The purpose of this segment of the program is to focus the development of component technologies into applications and environments which will demonstrate their utility and additional capability when incorporated into operational systems. These technologies include virtual reality telepresence, advanced display technologies, proximity sensing for perception technologies, and robotic flaw detection. The target applications include such tasks as repair of free-flying small satellites, and ground-based control of robotic servicers. Each of these areas have been identified by the potential space robotics user community as applications where space robotics will be necessary to satisfy their planned requirements. This user community includes the External Work System and anticipated commercial space system developers.

Ranger Telerobotic Flight Experiment

Telepresence / VR Control of Free Flying Robots

Portable Telepresence / VR Control Station

Ranger Task Control and Visual Tracking

Dexterous Arm Control for Ranger Flight Experiment

Space Operations

Technology Transfer Roadmap Details

Ranger Telerobotic Flight Experiment

In June 1992, the decision was made to actively pursue the development of the Ranger Telerobotic Flight Experiment (TFX), as proposed by the University of Maryland Space Systems Laboratory. This project includes the development of neutral buoyancy and flight prototypes for a class of low-cost expendable telerobots designed for research and servicing in space, beyond the space station orbit. The vehicle will be equipped with four manipulators: two 7-DOF arms for bilateral dexterous manipulation; a 7-DOF manipulator for grappling at the local worksite; and a 6-DOF arm for positioning a pair of stereo video cameras giving primary feedback to the remote operator. A second stereo camera pair mounted on the vehicle centerline will provide a stable visual reference for free-flight maneuvering, and ultimately feed a vision system for autonomous vehicle docking. Much of the design and construction of the Ranger Neutral Buoyancy Vehicle was completed in FY 93. In FY 94, the manipulators were assembled and integrated onto the completed mobility base. The completed Ranger NB vehicle, while developing telerobotic operational data, will be used to develop and verify algorithms, software, and experiment designs for the space flight experiment.

Focus and Directions:

FY94 Complete and test the neutral buoyancy version of Ranger; complete Phase A and B studies on the Ranger Telerobotic Flight Experiment (TFX)

FY95 Collect data base on advanced operations using neutral buoyancy vehicle; develop and test flight article

FY96 Fly the Ranger TFX; correlate in-space operations to neutral buoyancy simulation

Additional information and specifics associated with the focus and direction of the flight experiment can be found in the Preliminary Design Review documentation.

Point of Contact:
Dave Akin
(301) 405-1138
dakin@ssl.umd.edu

Telepresence / VR Control of Free-Flying Robots

The objective of this research task is to add a telepresence/virtual reality control interface to a free-flying robot. This interface will initially provide advanced teleoperation and supervisory control capability to the Ranger NB (Space Systems Laboratory, Univ. of Md.) by augmenting existing control stations. Subsequent work will provide telepresence/virtual reality control capability to the Ranger TFX telerobotic flight experiment.

The Ranger NB vehicle is an extremely complex telerobot, offering 32 degrees-of-freedom for operator control. To productively use this system, it will be critical for operators to have an effective control interface. In particular, the interface must provide support to the operator for visualizing the workspace, for efficiently displaying vehicle state, and for handling system latencies such as communications delay. In addition, the interface must enable higher human performance while reducing operator fatigue and stress. One approach which appears to satisfy these requirements is the telepresence/virtual reality control interface.

During the past three years, the Intelligent Mechanisms Group (IMG) has been developing control interfaces utilizing telepresence and virtual reality technology. These interfaces seek to provide robotic systems with high-fidelity telepresence capabilities and allow users to more easily interact with remote devices. By utilizing real-time interactive computer graphics, stereoscopic video and stereoscopic displays, telepresence/virtual reality interfaces enable users to efficiently manage and visualize complex systems. As a result, such interfaces can dramatically improve human teleoperation performance, particularly in the presence of time-delays.

Approach:

Our approach is to provide for a transfer of control interface technology from the Ames' IMG to the U.Md. SSL. This transfer will involve the augmentation of existing SSL operator stations with telepresence and virtual environment subsystems. Specifically, the IMG will provide technology developed at ARC which utilizes real-time interactive computer graphics and stereoscopic video displays. The work will be conducted in a two phase project.

In the first phase, a telepresence/virtual reality interface will be developed for the Ranger NB system. This interface will provide the operator with real-time visualization of the Ranger NB system state and worksite. The focus of this phase will be to directly enhance the capability and to improve the performance of human operators in a research environment.

In the second phase, refinements to the telepresence/virtual reality interface will be developed to support the Ranger TFX flight experiment. These refinements will include the development of orbital dynamic vehicle models and predictive displays for handling system latencies in the presence of communications delays. We expect that the control interface project will have a minimal impact on the SSL's on-going Ranger NB and Ranger TFX development process. The initial work will be performed at ARC and leverage existing IMG resources and personnel. This will be followed by integration with Ranger subsystems, which will be conducted jointly by the IMG and the SSL.

Focus and Directions:

Oct 93 Installed Telepresence/VR Control Station Hardware at ARC

Aug 94 Completed Integration of Telepresence/VR Control Station Software at ARC

Oct 94 Installation of Prototype Telepresence/VR Control Station Software at SSL for Ranger NB Roll-Out

Jan 95 Perform NB Human Factors Study of Traditional vs. VR Control Station (at MSFC Tank? TBD)

Sep 95 Completion of Ranger TFX dynamic simulation model integrated into VR Control Station

Early CY97 Operator control station for Ranger during the flight experiment

Early CY97 Network observer station for Ranger during the flight experiment

Point of Contact:
Laurent Piguet
(415) 604-6063

Portable Telepresence / VR Control Station

The objective of this task is to produce a prototype of a portable telepresence/virtual reality control station suitable for locally operating robotic vehicles and manipulators. The design target for the portable station is a briefcase-sized, lightweight, battery-powered system with the ability to operate the entire Virtual Environment Vehicle Interface (VEVI) control system. The system should achieve at least 80% of the performance of the current ARC control system, which uses a Silicon Graphics Skywriter and is currently the highest performance VR rendering platform at NASA Ames.

Virtual environment and telepresence techniques together form a very powerful operator interface for complex robotic systems. Over the last three years, the Intelligent Mechanisms Group at NASA Ames has developed the Virtual Environment Vehicle Interface (VEVI), which has been used to control various vehicles, including the Antarctic TROV submersible over a satellite link from the U.S., and the Russian Marsokhod rover in Moscow. VEVI is designed to allow an operator to use a combination of telepresence and virtual environment paradigms, the choice of which depends on the bandwidth and time delay of the communications link to the robotic vehicle. To date, the VEVI control system has been implemented on mid-to-high end Silicon Graphics workstations to generate the interactive 3D graphics upon which the virtual environment is based. This means that control interface is usually situated at a dedicated control center, and a satellite link is necessary for the command/data link to the vehicle. While this approach can serve a wide variety of large missions, it does not allow for easy field deployment and use of the operator interface near the work site.

Recent advances in processor design have produced architectures such as the i486, i860, and the Pentium, which provide an incredible amount of processing power in a very small package. Performance tests on these systems indicate that they are able to provide nearly equivalent rendering speed to mid-range Silicon Graphics workstations for specific applications. The advantage of these systems, however, is that there are strong commercial forces external to NASA pushing these architectures into small portable packages. We believe it is now possible to produce a briefcase-sized battery operated computer system powerful enough to support the full VEVI control software. Such a system would not only be used for NASA development and testing of telerobotic systems, but could be commercialized and sold for use in terrestrial applications. We have an industry partner, Sense8 Corporation, interested in producing a commercial product line based on the VEVI control software running on a portable workstation.

This task would not seek to develop new telerobotics or virtual reality operator interfaces, but to port an existing set of control software to a new hardware package. As such, it is a fast one-year effort, and should result in a complete and operational prototype system which could be used immediately for field operations.

Approach:

Our approach is to have the industry partner, Sense8 Corporation, provide an integrated portable computer system based on either an i486 or a Pentium processor, along with a pair of i860-based graphics accelerators for the VE stereo rendering. We (at ARC) would add video input and stereo decoding capability to the hardware. The system would be self-contained, and would include a head-mounted head-tracked display, a 6 DOF input device, RS-170 video input and stereo decoding capability for telepresence support, and network and telemetry I/O channels. NASA Ames would port the VEVI control system software onto this architecture using an existing version of the World ToolKit simulation software for the i486/i860 combination. We would benchmark the portable system against our existing high end control station which uses a four processor Silicon Graphics Skywriter, to determine its performance level. We would then deploy the portable system in a field test, using any of a number of available vehicles, to verify its performance in the field. Once the system has been verified operational, other TRIWG participants could acquire identical systems from Sense8 Corporation for their own applications.

Milestones:

Oct 94 Initiate Procurement of workstation hardware

Nov 94 Begin port of VEVI software to i486 architecture

June 95 Receive workstation hardware from industry partner

Aug95 Integrate hardware/software system and lab test

Sep 95 Field test portable control system

Point of Contact:
Butler Hine
(415) 604-4379
hine@ptolemy.arc.nasa.gov

Ranger Task Control and Visual Tracking

In FY 1995, this task was incorporated into the Ranger Telerobotic Flight Experiment task.

Point of Contact:
Dave Akin
(301) 405-1138
dakin@ssl.umd.edu

Dexterous Arm Control for Ranger Flight Experiment

The goal of this task is to augment the operational capabilities of the dexterous arms in the Ranger Flight Experiment. These new capabilities will be developed within the framework of Configuration Control, which has been developed at JPL and selected for implementation on the Ranger arms. The specific objectives of the task are as follows:

Develop the capability of on-line collision detection and avoidance for the Ranger dexterous arms. This capability does not currently exist in the Ranger baseline control system, and erroneous operator commands can cause collision between the dexterous arms and the camera and grapple arms, the base, or the task board. The performance improvement due to this added capability will be measured in two ways. First, it will enable a broader range of tasks to be executed safely, such as collision-free reach inside a constricted space or opening. Second, it will cause a reduction in the Ranger operation time by 50%, since several possible motions with potential collision are not executed. Finally, this capability will increase the safety of the Ranger during the operation of the arms, a feature which is vital to the success of the Ranger mission.

Provide the ground operator a software tool for proper placement of the Ranger base. This algorithm will ensure that both dexterous arms reach the task site and the useful workspace volume is maximized. The algorithm will take into account the fact that the Ranger base is attached to the vehicle by the grapple arm. At present, the placement of the Ranger base is done by the ground operator in an iterative trial-and-error fashion. The performance improvement due to this added capability is expected to be a reduction by 30% of the Ranger operation time.

Develop cooperative control schemes for the two Ranger dexterous arms to perform dual-arm tasks such as satellite servicing operations. In particular, the arm redundancy will be exploited to minimize the torque imparted on the base during the dexterous arm motions. The performance of the Ranger base with and without this capability will be demonstrated and compared. The baseline performance will be documented and the improvement will be quantified.

These three capabilities will considerably enhance the robustness and reliability of the Ranger arm control system, and will significantly expand the range of tasks that can be accomplished in the Ranger Flight Experiment. A series of technology experiments will be conducted both at JPL and SSL to demonstrate control of dexterous 7-DOF arms in various control modes, from direct teleoperation to supervisory control.

Focus and Directions:

FY '95 Develop and implement collision detection/avoidance and base placement capabilities for the Ranger to enable robust and reliable task execution. Perform proof-of-concept graphical simulations on the Silicon Graphics IRIS Workstation at JPL using the kinematic model of the Ranger 7-DOF arms and the task site. Transfer the JPL-developed software modules to the Ranger Flight Computer and the Ground Control Computer.

FY '96 Demonstrate dexterous cooperative dual-arm control at JPL using the existing two 7-DOF Robotics Research arms. Conduct experiments using teleoperated and supervisory control modes for various tasks. Transfer the JPL-developed software modules to the Ranger Flight Computer.

Point of Contact:
Homayoun Seraji
(818)354-4839
seraji@telerobotics.jpl.nasa.gov

Space Operations

The development of new technologies for space telerobotics brings with it a concomitant requirement for understanding the impact of that technology on the operational capabilities of the eventual telerobotic system. Addressing this area of space telerobotic operations is the primary focus of the University of Maryland Space Systems Laboratory (SSL). During its initial years of operation at the Massachusetts Institute of Technology, the SSL pioneered the development of analytical models for neutral buoyancy simulation, and performed extensive tests on extraÐvehicular operations, leading to the Experimental Assembly of Structures in EVA (EASE) tests on STS 61-B in late 1985.

Since that time, the SSL has focused primarily on space telerobotic operations, with emphasis on neutral buoyancy simulations of integrated EVA/telerobotic work sites. The Beam Assembly Teleoperator (BAT) has performed structural assembly of both EASE and Space Station truss structures, as well as tests of Hubble Space Telescope servicing, both alone and in conjunction with EVA subjects. The Multimode Proximity Operations Device (MPOD) has performed a number of tasks relevant for orbital maneuvering vehicleÐclass spacecraft, and has demonstrated the utility of manned astronaut support vehicles for extended EVA capabilities. The Apparatus for Space TeleRobotics Operations (ASTRO) has been used to research three-dimensional positioning and station keeping systems. The Stewart Platform Augmented Manipulator (SPAM) replicates the functionality of the Space Shuttle Remote Manipulator System, with improvements in fine end-point positioning based on the Stewart Platform wrist. The Supplemental Camera and Maneuvering Platform (SCAMP) provides operator-controllable external video views, and has been used for tests of single-operator control of multiple free-flying telerobots.

Focus and Directions:

FY94 Collect data base on advanced telerobotic operations using neutral buoyancy; develop an advanced work site simulation for quantifying performance of integrated telerobotic operations; test EVA/telerobotic cooperative tasks at NASA Marshall Neutral Buoyancy Simulator

FY95 Operate existing telerobotic systems to collect data base on advanced telerobotic operations in neutral buoyancy; develop an advanced work site simulation for integrated telerobotic operations; test EVA/telerobotic cooperative tasks at NASA Marshall Neutral Buoyancy Simulator

FY96 Utilize Ranger technology for rapid prototyping and operations testing of advances concepts for telerobotic and EVA/telerobotic space operations

FY97 Use results from Ranger flight experiment and Ranger NBV to develop extensive data base on telerobotic performance in space operations tasks

Point of Contact:
Dave Akin
(301) 405-1138
dakin@ssl.umd.edu