Robot Motion Planning with Trajectory Wide Constraints

Project Description At Carnegie Mellon University’s Personal Robotics Lab, we aim to create algorithms that enable robots to seamlessly and autonomously collaborate with and around people in human-centric environments, such as the home. One of the many challenges in creating this future is giving our robots the skill set to perform many of the tasks we do on a daily basis, from the simple to the complex. Many of the tasks that we do have context-based constraints. For example, if you wanted to carry an empty coffee mug from the counter top to your desk there are very few constraints on how you hold your mug, as long as it is a relatively stable grasp. Many manipulator robots, including our lab’s HERB (Home Exploring Robot Butler) [1], could perform this same task. However let’s re-imagine the scenario with the coffee mug being full of coffee. Now transporting the mug has the additional constraint of keeping the mug upright in order to not spill the coffee. While we human figure this out easily, we need to be able to endow our robots with this same kind of contextual knowledge. Ideally we want to be able to define this constraint, and many other possible constraints, easily, and pass these constraints on to our existing motion planners. This would allow our robot to create plans that adhere to the context’s constraints in the same way that humans do.