|
|
|
THE TEAMBOTICA PROJECT |
|
|
 |
|
|
TEAMBOTICA is a research environment for the
exploration of theories, designs and implementations of team-based
robotics.
"A bioterrorism alert is
recieved in a dense urban environment. A team of autonomous air
and ground vehicles is dispatched. The team enters the
environment, assesses the situation and collaboratively
coordinates its activity. A biosensor detects toxic gases, the
team dynamically positions itself to map out the plume of gas
and follow its progress."
A new software capability for the reprogrammable, coordinated command and
control of teams of autonomous unmanned air and ground vehicles (UAVs/UGVs)
based on an integrated planning and distributed control architecture in
which collective behavior is uniquely adaptive and fault tolerant. This
work extends the Multi-level agent architecture (MLAA) developed under
a previous ONR contract.
The key research objectives are to develop and demonstrate :
-
Interaction methods through which the mission planning level within the
MLAA can negotiate with the networking layer, trading mission-level requirements
for suggested network resource allocations and expected networking constraints
for supporting robot localizations
-
A decision-theoretic approach in which desirabilities/utilities are
assigned to objectives relative to localization, navigation and mission
constraints on a robot's choice of task
-
Probabilistic methods for projecting the current state in the context of
prevailing constraints and a team's uncertain information about the current
situation
Tangible Benefits of TEAMBOTICA
-
An adaptive, robust architecture for collaborative agents
-
Collaborative perception-based localization
-
Vision-based algorithms to support remote positioning
-
Negotiation protocols to support collaboration across multiple dimensions
(e.g. communication, localization and mission)
-
Temporal projection to assess the impact of an action at different
levels of abstraction
-
A general framework for execution monitoring
|
|
