Task Planning

Robotics planning problems exhibit strong geometric constraints that, at the high level, affect what can be done. For example, before deciding to pick up a cup, we should determine whether it is geometrically feasible to move the robot base to an appropriate position. One major research thread is therefore the integration of task and motion planning.

Even the best-laid plans often go awry. A robot must therefore execute its plans in a robust manner, reacting quickly to minor disturbances, and re-planning when faced with major unexpected changes. For this purpose, we use TREX, a sophisticated executive based on temporal logic and constraint programming. An ongoing research focus is making TREX more accessible to users without experience in logic-based specifications.

Task-level planning requires information about the objects in the world and their relationship to one another. Task-level planning also requires awareness of what is uncertain or unknown, allowing the robot the opportunity to take further sensing actions. An area of interest is therefore to build and make use of higher level representations of the state of the world.

Relevant Blog Posts

Publications

2010
Combined Task and Motion Planning for Mobile Manipulation Wolfe, Jason., Marthi, Bhaskara., and Russell, Stuart International Conference on Automated Planning and Scheduling, 05/2010, Toronto, Canada, (2010)  Download: ctamp-icaps-2010.pdf (134.6 KB); ctamp-tr.pdf (188.95 KB)
2009