Motion planning in an dynamic environment is complicated by the need for constant adjustments of plans to account for stochastically moving obstacles. Yet, it is critical in real-world applications such as flight coordination, assistive robots and autonomous vehicles. We focus on developing real-time motion planners for environments with a large number of stochastically moving obstacles and robot sensor uncertainty. We developed methods that works even when moving obstacles have strongly-interacting stochastic dynamics or completely unknown dynamics.

Robotic task learning is often difficult to achieve due to the complex nature of the tasks and the complex dynamics of the robot. We explore the application of reinforcement learning (RL) in order to learn complex motions for robotic task. Specifically, we explore new methodologies in order to operate in complex continuous robotic task spaces and optimize the time to learning convergence.

IgE antibodies bound to cell-surface receptors, FceRI, crosslink through the binding of antigens on cell surfaces. This formation of aggregates is what simulates mast cells and basophils in order to initiate an allergic response. Experimental studies have shown that the spatial organization of aggregated IgE-FceRI complexes affect transmembrane signaling that initiate these responses. Using motion-planning inspired methods, we simulate the interactions of hundreds of molecules.

Kinesin-1 is a microtubule motor protein which is responsible for the transport of vesicles along axons. Impaired motor protein axonal cargo transport may be implicated in a variety of human neurodegenerative diseases. We investigate the kinesin stepping mechanism by combining robotics techniques with energy calculations of 3D geometric models of the kinesin head and microtubule. The goal of this computational study is to enable a mechanistic understanding of how molecular scale obstacles on the microtubule track affect kinesin motion.

Effective collaboration betweens humans and automated systems can enable performance simply not possible fully automated or fully human control, especially in dynamic and uncertain environments.

Molecular docking is an important problem in biology for the study of immune systems, allergies, and many more processes. Automated methods for solving molecular docking can be computationally expensive, instead, we use an approach to take advantage of human intuition. To accomplish this, we developed a molecular docking game that was expanded to a crowdsourcing application.