Introduction: Over the past seven years, veterans with high level quadriplegia have been testing the ability of a robotic workstation to replace an attendant for independent execution of vocational tasks. These tests have proven feasibility, and along with the recent VA Technology Transfer Section (TTS) evaluation, have clarified the next necessary phase of development that will advance the workstation to successful commercialization. The primary goals for this phase are more functionality per dollar, easier operator control and higher system reliability. With recent commercial and academic research advances in technology, we believe that we can realize these goals in the framework of a development project in the next three years.

Improved Functionality: We will refine the current DeVAR interface so that users will be able to design their own robot tasks without having to become robot programmers to do so. The graphical interface RoboGlyph, based on a commercial visual communication system, will use icons to represent robot actions and allow graphic visualization of the robotÕs motion and sensor information. RoboGlyph will replace the current interface, which is based on a classical, text-based robot control language and rudimentary voice control. In the new system, increased functionality will be achieved with force and touch sensors that detect surfaces and objects, allow the robot to make adjustments when needed, and verify that tasks are done correctly and safely. The combination of the new interface and the sensors is the key to a successful implementation, since neither advance alone will meet the challenge of providing effective user control of robot programming.

Enhanced System Reliability: The reliability of the robot will be increased by a commercial control system that integrates motion, sensing and communication functions. The controller will be networked to the interface computer, with a link to the Internet. This modular approach will enhance reliability, and will provide a rich, multimedia communication path for troubleshooting, training and information exchange between the user, the robot manufacturer, therapists and other users. The TTS evaluation demonstrated the tremendous importance of communication to system maintenance and downtime reduction. We will customize Stanford-developed, network-based software for remote diagnostics and troubleshooting that will allow for fast response to problems that arise.

Cost-Effective System Design: The new, networked system design approach will enhance the usability, effectiveness and functionality of the robot without increasing cost. The system will provide therapists and users with network-based tools they need for robot-related functions, i.e., e-mail, the World Wide Web and other on-line services. Just as they are essential for the robot, these tools can also enhance overall clinical productivity. Other desktop functions, such as database management and word-processing, can be added at very little additional cost, making the resulting robot-equipped work environment a centralized resource for both clinicians and patients. Cost effectiveness will also be achieved because only one computer interface needs to be customized per user, eliminating the need for dedicated interfaces for each additional daily living device the user might want to operate.

Clinical Testing and Performance Analysis: The interface, controller and network system design will be integrated into a single advanced workstation that we have designated ProVAR. The advanced features of ProVAR will be tested initially by performance analyses and on the basis of patient and therapist feedback about the effectiveness of the system for manipulation, communication and therapy training. Final clinical evaluation of the workstation, refined on the basis of the initial phase of testing, will be performed by incorporating the use of ProVAR into the daily treatment program of the Palo Alto VA Spinal Cord Injury Center.

Commercialization: Technology transfer to industry will be pursued through early and ongoing interactions with two corporations that are already interested in a ProVAR product. We will continue to seek other corporate partners during the grant period, and seek TTS involvement in any subsequent evaluation of ProVAR.

Status (12/19/96): A ProVAR website has been established (Check it out!).
The staffing is nearing completion. The interface design work is being spearheaded by Joe Wagner, and the robot controller software aspects by K.C. Chang, both Stanford University doctoral students. Neils Smaby, Master's student in the Stanford Design Division, is focusing this year on proximity sensor development for the robot arm. Professor Jun Ota from the University of Tokyo, visiting Stanford for the 1996-1997 academic year on sabbatical leave, has an extensive background in robotics and will be assisting the project on interface design. Prof. Larry Leifer of the Design Division, Dept. of Mechanical Engineering, and Prof. Oussama Khatib, Computer Science Dept., director of the C.S. Robotics Lab, are active advisors to this project.