Learning for Control of Synthetic and Cyborg Insects

Abstract:

We aim at enabling operation of synthetic and cyborg insects in complicated envi- ronments such as outdoors or inside collapsed buildings. Success in this research project will bring society closer to solving the grand challenge of having teams of mobile, disposable, search and rescue robots which can robustly locomote through uncertain and novel environments, finding survivors in disaster situations, while removing risk from rescuers. Learning and adaptation capabilities are critical to handle significant uncertainty in mobile platforms dynamics and environment vari- ables. As part of our cyborg insects thrust, we developed a flexible, multi-electrode polymer interfaces used to construct stable, chronic interfaces to the sensory organs of adult insects, and showed that the implanted interfaces can chronically record stimulus-triggered evoked potentials from eye and antennae. We are currently in- vestigating second generation interfaces based on new materials: carbon fiber with PEDOT and parylene with platinum electrodes. We have also been investigating a new locust steering device based on contrast stimulation. As part of our synthetic insects thrust, we instrumented an experiment are with an OptiTrack motion cap- ture system to investigate the gait properties of our new VelociRoACH crawler. We found that reinforcement learning methods can improve the robot's speed on different surfaces. Additionally we have conducted preliminary experiments on visually guided flight of synthetic flying insects. As part of our thrust towards collaborative exploration algorithms, we developed a new method for learning to control dynamical systems (including both synthetic and cyborg insects) starting with almost no prior information about system dynamics. Our method is based on a provably efficient exploration technique that we have extended from discrete to continuous state space. Preliminary simulations show fast learning with minimal supervision.