Planning involves converting high-level objectives for a robot or robot team into specific sequences of actions that can be executed using lower level control algorithms tailored to the individual robots.
The complexity of planning in heterogeneous robotic deployments necessitates abstraction and decomposition of planning tasks into three components:
- High-level mission planners,
- Motion planners for defining safe motions that achieve each task in the mission, and
- Low-level controllers for executing motion plans as sequences of motion primitives for which trajectory tracking or path following controls can be employed.
The high-level mission planning methods must be defined in such a way that they remain consistent with low-level robot motion limitations, and yet must remain sufficiently abstract to result in real-time adaptation or re-planning capability when new information and events affect the optimality of the current plan.
Within this research theme, we will focus primarily on three topics:
- Multi-robot task allocation, building on Professor Smith's work on Large Neighbourhood Search solvers for task allocation (as part of a collaboration with Clearpath Robotics);
- Coordinated motion/action planning, building on Professor Smith and Professor Waslander's work on persistent aerial surveillance and aerial package delivery with teams of flying and ground robots; and
- Developing a unifying language for motion tasks, building on Professor Smith's work on rigorously expressing logical constraints in robot path planning problems by combining the satisfiability and traveling salesman problems.