Simulation and Virtual Environments
We are interested in developing simulators and virtual environments for multiple purposes. Traditionally, simulators have been used as replacements for the real world when testing robot algorithms and mechanism. However, combined with new, generally available powerful computing resources, simulators can be hugely valuable beyond this role. They can be used as a computational backend for the real world, testing multiple possible action primitives and scenarios during a planning loop for a real robot performing a task. Equally interesting, simulation can be an integral part of mechanism design, running optimization algorithms over numerous scenarios that the robot being designed is expected to encounter. We are interested in many aspects of simulation and virtual environments, from collision and contact detection to rigid- and soft-body physics and to rendering and interfacing with the user.
We are using and actively developing the GraspIt! simulator for robotic grasping, started at Columbia University and now used by numerous research labs around the world.
Many of our projects using simulation are under current development. We will post highlights here as they become available. Examples of using virtual environments for robotic manipulation research can be found in some of our previous work.