The current state of our project, described in Section 3 below, makes ten mobile robots available for experimentation over the Internet. This is of interest given recent levels of activity in cooperative mobile robotics . We provide the remote researcher with (i) access to a colony of near-leading edge research robots, (ii) software and hardware extensions to enable wireless networking and programmability in high-level languages, and (iii) infrastructure for remote experimentation. Thus, our project truly essays experimental remote science.
For present lessons to be more applicable to future efforts, and for relevance to ``real science'', we have centered on themes of mobility, autonomy, and remoteness in evaluating potential testbeds. Currently, we anticipate two future project phases which lead to unstructured outdoor environments and applications such as remote exploration, geological and botanical sampling, and ecosystem monitoring. The three project phases, and their relationships to each other, may be summarized below.
Phase One (in progress) provides an indoor experimental facility allowing users from cooperating institutions to log in and control our robotic colony via a World-Wide Web (WWW, or Web) interface. Here, we believe users will focus on research issues in cooperative mobile robotics. Initial functionality includes a full Unix development environment (X, gcc, Emacs, gdb, etc.), Web interface, simple scripting capability, robot control via a shared-memory architecture, and full Internet connectivity (TCP/IP, SMTP, FTP, etc.).
Phase Two will introduce behaviors (i.e., task primitives, high-level tasks) into the colony via a layered control approach. Of highest priority are behaviors which preserve the safety and integrity of the colony (see  as well as the discussion of requirements and implementation issues in the remainder of this section). This ``safety net'' will provide protection against errant or harmful user programs, against hardware failures and power loss, etc. At the same time, the robot arena will be enhanced with hardware to allow manipulation of large obstacles, thus affording more complex experimental environments and the removal of failed robots. Given this increased flexibility and programmability, the sophistication of experiment design will increase; we will provide more extensive data logging (sensing, actuation, variable-resolution or filtered on-board video), and experiment profiles. Finally, the interface will improve, following the development of HTML3 and other Web facilities.
Phase Three will field equipment in an outdoor setting, using heterogeneous (legged or possibly aerial) robot platforms which are almost fully autonomous. Experiments will last on the order of days to weeks, and will be run remotely over the Internet. Science (e.g., botany, soil science, and atmospheric chemistry) will be performed, with task complexity hopefully on the order of that achievable by the JPL Pathfinder Mars rover . We believe that an important benefit will be near-continuous, close-in monitoring for regions in which human access is too costly.
Throughout, our system will support rigorous scientific experimentation by enforcing well-defined experimental protocols via its interface. Note also that our Phase Three does not entail teleoperation or telepresence; such approaches are inappropriate for reasons including network failure and the high cost of continuous human operation. Rather, our goal is to extend the robots' sensing and manipulation capabilities to optimize information gathering and delivery to the user. The remainder of this section lists functional requirements for our system and contrasts them with capabilities afforded by existing efforts.