Precision control technologies for ROVs and intervention AUVs
Lead Engineer/Project Manager: Steve Rock (Stanford University)

We propose to continue this program into its second year. We are on-track in meeting the objectives set for 1999, and the tasks originally proposed for 2000 are still appropriate. In particular,

• We have made excellent progress in updating the underlying technologies required for vision-based navigation and control of an ROV or AUV (including the validation of portions of the new software which implements these algorithms on Ventana).
• We have also made excellent progress towards the goal of demonstrating precision manipulation involving combined vehicle/manipulator control of a two-link arm mounted on the OTTER vehicle. Models of the hydrodynamic loads have been developed and validated, and algorithms to manage these loads within a control system have been conceived.
• We have begun to explore techniques to incorporate end-point vision-based feedback to aid a pilot in operating a robotic arm mounted on an ROV (e.g. the ISE manipulator on Ventana) 

If this coming year we propose to continue the above activities, we also propose

• To explore the extension of the vision system to include 3-D information of the scene. This will be useful both for the display of data gathered to a scientist, and to the improved navigation and control of the ROV/AUV.
• To explore the fusion of other sensors (with vision) to improve the navigation and control capabilities of and ROV/AUV.

There are several issues that need to be addressed in the preparation of this proposal:

• The thrusters and power system on the OTTER vehicle have proven to be problematic. Since OTTER is required for the completion of tests currently in progress, we need to address the issue of upgrading this system to improve both its safety and utility.
• We need to coordinate our future needs for AUV test and development platforms with other emerging activities at MBARI. In particular, it is likely that Odyssey vehicles will exist. Commonality of platforms is desirable if this can be done without compromising research objectives.
• We need to reevaluate our immediate and long term research objectives both to support and to exploit the results of other MBARI research programs (e.g. in support of the MOOS). The goal of this effort is to explore high-risk high-payoff technologies that will enable a "spectrum of autonomy" in unmanned underwater vehicle control. At one end of the spectrum are pilot aids for ROV operation (e.g. automatic station keeping, automatic mosaic generation, point-and-click manipulator control, etc). At the other end of the spectrum is fully autonomous mission capability for intervention capable AUVs and sub-missions on ROVs (albeit for mission tasks sufficiently simple not to require the full creativity of a human pilot). The immediate focus of our research will be in two technology areas. The first is the development and demonstration of automatic vision-based feedback control (i.e. closing the loop through a computer instead of the pilot) as a means of controlling ROVs, AUVs and manipulators. The second is the development of techniques for performing high-performance task-level (e.g. point and click) control of manipulators operating from either a fixed base (e.g. an ROV sitting on the bottom) or from a free-floating small AUV.

An additional goal of this program is to provide a path through which scientists, engineers and pilots at MBARI can affect Ph.D. level research at Stanford University so that new technologies of interest to MBARI (but that ate considered too high-risk for development) can be explored. It will also allow new concepts developed at Stanford to be implemented, tested and evaluated by interested scientists, engineers and pilots at MBARI.