The next frontier for robotics is bringing robots into environments such as homes, flexible manufacturing, and disaster sites, where the standard robot design, consisting of rigid links actuated by electric motors, limits their usefulness and creates safety concerns. Unlike rigid robots, soft robots have the potential to be highly accessible, versatile, and inherently safe around humans. However, there is currently a lack of engineering principles for designing soft robots that perform practical tasks. It is also unclear how to enable these robots to move reliably and efficiently in cluttered environments, like a disaster site. This work explores new methods for designing, controlling, and planning motion for soft robot arms.
The technical objective is to explore, formalize, and validate the fundamental algorithmic and engineering principles necessary to build soft robot arms and allow them to perform useful tasks. To that end, the project investigates (1) A modular soft arm design where each module can offer a combination of 3D bending, extension, and change in stiffness, (2) Modeling and control methods to achieve reliable position, force, and impedance control, and (3) Motion planning algorithms which plan in the belief space of the robot and exploit contact with the environment to overcome uncertainty in the motion model.
