Develop software control system for autonomous
mobile robots that:
-
accept mission-level commands
-
learn from experience to use / acquire
behaviors
-
can be trained with intuitive interface
-
share learned knowledge with other robots
Project Approach
During the first year, a multi-agent based Human-Robot Interface (HRI)
approached was designed and the following key components were developed:
the Self Agent, a compound agent responsible for monitoring the status
of the robot itself and communication with the human commander and
other robots; the Sensory EgoSphere (SES), a short-term memory data
structure for dynamically storing sensor data; the Database Associative
Memory (DBAM), a long-term memory in which records are linked through
a spreading activation network (SAN).
During the second year, various behavior agents were developed
for mobile robots including the ATRV-Jr, Pioneer-2-AT, HelpMate and Trilobot robots. A user-centric GUI for the ATRV-Jr and the HelpMate was developed.
A mission planner was developed and will be integrated into the
GUI. We design and implemented Sensory EgoSphere (SES)- and Landmark EgoSphere
(LES)-based navigation and knowledge sharing using the ATRV-Jr
and Pioneer2 robots. Robot learning was developed and tested
on the ATRV-Jr using reinforcement learning algorithm.
