Autonomous robots in unengineered or partially engineered environments
Much of the work in mobile robots has been based in engineered
environments. By this I mean an environment which is well known, both
in terms of the physical layout and in the events which are allowed to
occur. However, many potential applications for robotics exist in
unengineered or partially engineered environments. These environments
place additional requirements on the robotic systems, especially if
they need to operate in autonomous, or shared control modes.
Many of the current initiatives in robotics require robotic systems
that can function in either completely unengineered environments, or
environments which were once engineered but have undergone
un-documented changes. For example the current Robotics and
Intelligent Machines initiative by the Department of Energy (RIM) Roadmap calls out several
Technology Drivers which require the ability to function in these
environments.
Given the growing need for these capabilities, we are proposing
research into the following key area: Light-weight planning under
uncertainty.
What is an engineered environment?
What additional demands are created by unengineered
environments?
What additional capabilities are needed by robots that function in
these environments?
How would Probability-Aware Planning help?
The ability to function autonomously, in the face of sensor
uncertainty, operator uncertainty, and exogenous events is required in
many of the key technologies called out by the RIM Roadmap. In
specific these are required by:
- Autonomous RIM for Hostile situations, hazardous releases
(Reduce Hazards Epoch II)
- Autonomous rover RIMs for monitoring, etc. (DP Reduce Costs,
Epoch II)
- Automated Decon of operation facilities (MD: Exposure Reduction,
Epoch I, II)
- Automated Welding/Cutting NE: Improve Operations, Epoch II)
- Automated handling, painting, cleaning (NE: Maintenance, Epoch
I)
- Automated navigation path planning (EM: Exposure Epoch III)
- Autonomous/remote surface survey (EM: Waste Reduction Epoch III)
- Autonomous action selection, map generation (ER: Missions Epoch
I, II, III)
jpg3u@virginia.edu
Last modified: Tue Dec 19 14:57:15 2000