Thesis Proposal
share
In this thesis, we will attempt to develop new algorithms that enable an agent to deal with uncertainty in application to planning of robot grasps and manipulative actions. We focus on constructing a system where an agent interacts with its surrounding environment by selecting the most appropriate action at any time in a task-oriented fashion. Our main innovation is to split the planning problem into: (1) an RRT-based planner operating in the configuration space of the object, and (2) a local planner which generates a sequence of actions in the joint space that will move the object between two pairs of nodes in the RRT.
READ FULL TEXT