Simultaneous Motion and Task Planning Using Task Reachability Graphs
Advisor Information
Raj Dasgupta
Location
UNO Criss Library, Room 232
Presentation Type
Oral Presentation
Start Date
7-3-2014 2:45 PM
End Date
7-3-2014 3:00 PM
Abstract
Task and motion planning are two fundamental problems in robotics which are frequently encountered in many applications for robots. Task planning involves finding a sequence in a set of tasks satisfying some set of constraints. Path planning involves finding a path through the environment which is collision free between a start and goal location. Since, in most cases, a task involves going to a point in the environment to perform some operation, task and path planning are closely related. However, these two problems have been normally addressed as two separate research problems. Only recently researchers have considered combining these two topics using a technique called Simultaneous Task and Motion Planning (STAMP). In this research, I propose a new method to solve the STAMP problem using a framework called a task reachability graph (TRG). A novel feature of this approach is that it incorporates a very practical aspect of robotics - uncertainty in the robot's motion and uncertainty in the environment into the decisions made by the robot to determine the order of performing tasks while traversing the vertices of the TRG. I have validated the proposed algorithm using two Corobot robots performing different numbers of navigation tasks within an indoor environment. I have also compared it with another recent STAMP algorithm called MRTA-RTPP and shown that the TRG-based algorithm performs comparably.
Simultaneous Motion and Task Planning Using Task Reachability Graphs
UNO Criss Library, Room 232
Task and motion planning are two fundamental problems in robotics which are frequently encountered in many applications for robots. Task planning involves finding a sequence in a set of tasks satisfying some set of constraints. Path planning involves finding a path through the environment which is collision free between a start and goal location. Since, in most cases, a task involves going to a point in the environment to perform some operation, task and path planning are closely related. However, these two problems have been normally addressed as two separate research problems. Only recently researchers have considered combining these two topics using a technique called Simultaneous Task and Motion Planning (STAMP). In this research, I propose a new method to solve the STAMP problem using a framework called a task reachability graph (TRG). A novel feature of this approach is that it incorporates a very practical aspect of robotics - uncertainty in the robot's motion and uncertainty in the environment into the decisions made by the robot to determine the order of performing tasks while traversing the vertices of the TRG. I have validated the proposed algorithm using two Corobot robots performing different numbers of navigation tasks within an indoor environment. I have also compared it with another recent STAMP algorithm called MRTA-RTPP and shown that the TRG-based algorithm performs comparably.