UMass-LPR research on distributed control is aimed at creating short-lived virtual robots that share distributed resources to solve multiple, simultaneous perceptual and motor tasks. These tasks generally require the coordinated activity of several specialized sensory, mobility, and manipulation resources. For example, a virtual stereo vision system recruits a pair of cameras that can produce good quality information for localizing a particular subject. The pair can migrate over the set of available resources, or may recruit mobility resources to condition the imaging geometry in response to run-time feedback. In a virtual sensor, this coordination involves communication among sensory, motor, and control processes that are distributed over multiple processing sites.
Infrastructure
| Mobility Platforms | ||
|---|---|---|
 ATRV-Jr |
 UMass uBot |
 ATRV-Mini |
| Cameras | ||
 PAL Omnivision cameras |
 Sony PTZ EVID30 |
 Sony PTZ EVID100 |
The array includes dedicated computer resources for creating intelligent sensory and motor services that communicate through dedicated high speed networks. Lookup services and real-time support is provided by middleware including; JINI™, CORBA™, and NDDS™.
Application
Several platforms have been developed to experiment with implementations of robust distributed robot systems. Applications include sensor networks, coordinated teams of mobile platforms, attentional focus and tracking, and human-robot interfaces.
- Networks of Sensory and Motor Services
- Distributed Perception
- Coordinated Multi-Robot Control
- Interaction Based State
- Mixed-Initiative and Human-Robot Collaboration