Case Study
Friday, June 30
04:30 PM - 05:00 PM
Live in San Francisco
Less Details
This presentation will focus on the use of reinforcement learning, a type of machine learning approach, to teach autonomous agents how to make decisions that enable them to reach their destinations as quickly as possible. In order to achieve this, the agent must be able to interact with its environment and learn from the results of those interactions. While agents can be trained to be highly effective in simulations, they may struggle when it comes to real-world scenarios. We will explore the importance of trust and expectations in shared autonomy, specifically the interaction between humans and AI in autonomous vehicles. By collaborating with these factors, we can ensure a safe and efficient driving experience for all.
Matthew Walter is Associate Professor at the Toyota Technological Institute at Chicago (TTIC). Previously he worked as Research Scientist at Massachusetts Institute of Technology (MIT). His research focuses on advanced perception algorithms that endow robots with a rich awareness of their surroundings and the ability to interact safely and naturally with humans in unstructured environments. He is interested in algorithms that take as input multi-modal observations of a robot's surround (e.g., laser range data, image streams, and speech) and infer properties of the objects, places, people, and events that comprise a robot's environment, at a level of abstraction necessary to realize command and control mechanisms that are both intuitive and safe.
Matthew received his Ph.D. from the Joint Program between MIT and the Woods Hole Oceanographic Institution (WHOI) in relation to his research, which considered the problem of robot localization and mapping (i.e., SLAM) within large unknown environments.