See to Act, Act to See - The Interdependence of Planning and Perception in Robotics

Explore the interdependence of planning and perception in robotics through this 55-minute conference talk by Parker Ewen at Montreal Robotics. Delve into three key insights for improving robot interaction and learning from the environment: focusing on physical properties, utilizing multiple sensing modalities, and modeling uncertainty in sensor outputs and property estimates. Discover a novel approach for generating probabilistic models of a robot's environment using tactile and visual sensing, and learn how these models enable legged robots to accomplish tasks like locomotion more effectively. Gain valuable knowledge from Ewen's research in Bayesian inference, computer vision, and legged locomotion, drawing from his experience as a PhD student at the University of Michigan and his academic background from ETH Zurich and McGill University.