Professor, Sorbonne Université Director of the Institute of Intelligent Systems and Robotics
Title: Acquisition and Grounding of Robot Knowledge Through Interaction with the Physical World
Summary: Data-intensive approaches such as deep learning systems have a major issue: while they classify tremendous amounts of data, they do this without grasping the meaning of their inputs (data) or outputs (classes). It becomes easy to deceive such systems to falsely interpret some inputs. We contend that it is only though interacting with the real world that it is possible to make sense of it. Robots are machines that integrate perception and action, are situated and act in their environment. Perception should not be a one-way observation process isolated from action. Considering perception and action simultaneously enables to interpret the environment in terms of the robot's own perceptual and action capacities, thus relating the world to the robots potential activities. What an object affords to the robot, in terms of the robot’s action effects, partly describes its semantics and its functions. This grounds robot knowledge in the real world.
Bio: Raja Chatila, IEEE Fellow, is Professor of Robotics, Artificial Intelligence and Ethics at Sorbonne Université in Paris, France, and director of the Institute of Intelligent Systems and Robotics (ISIR). He also leads the SMART Laboratory of Excellence on Human-Machine Interactions. He contributed in several areas of Artificial Intelligence and autonomous and interactive Robotics, with about 150 published papers. He is recipient of the IEEE Robotics and Automation Pioneer Award. He is chair of the IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems. He was President of the IEEE Robotics and Automation Society for the term 2014-2015.
