Efficient collision avoidance task representation in a hierarchical manipulation framework

A stack of tasks formulation for robot control allows for imposing a hierarchy of desirable properties on the output behavior of the system. Tasks specified at higher priority levels are satisfied first, before moving on to refine solutions further and satisfy tasks of lower priority. Typically, a high-priority task is used to ensure that the robot does not collide with objects in the environment (or itself). Solving this task in a fast and non-overrestictive manner is therefore important for guaranteeing that lower-priority tasks can be satisfied in a timely manner.

This thesis will concentrate on implementing a collision avoidance task within a robot control framework. Generic geometric primitive models will be devised to represent the robot geometry. Different 3D space representations, constructed from on-board sensor readings, will be used to model the environment. In particular, collision avoidance based on occupancy grids and Truncated Signed Distance Fields will be considered and compared.

Contact: Robert Krug