Probabilistic Navigation in Partially Observable Environments
Reid Simmons and Sven Koenig,
School of Computer Science, Carnegie Mellon University Carnegie Mellon University
Autonomous mobile robots need very reliable navigation capabilities in order
to operate unattended for long periods of time. We have developed an
approach that uses partially observable Markov models to robustly track a
robot's location and integrates it with a planning and execution monitoring
approach that uses this information to control the robot's actions. The
approach explicitly maintains a probability distribution over the possible
locations of the robot, taking into account various sources of uncertainty,
including approximate knowledge of the environment, actuator uncertainty,
and sensor noise. A novel feature of our approach is its integration of
topological map information with approximate metric information. We
demonstrate the reliability of this approach, especially its ability to
smoothly recover from errors in sensing.