Preiss's research focuses on building rigorous and practical methods for learning-based planning and control in robotics and other autonomous systems. Key technical tools include online optimization, reinforcement learning, adaptive control, trajectory planning, and neural networks. He is also interested in developing theory to understand why robot learning can succeed in "messy" real-world settings where typical simplifying assumptions do not apply. James works with a variety of robot types including quadrotors, manipulators, and ground vehicles.