Colloquium | January 27 | 6:30-8 p.m. | Berkeley Art Museum and Pacific Film Archive, Osher Theater
Amy LaViers, Robotics, Automation, and Dance (RAD) Lab
Movement seems to encode information. How does this work? We know that animals, including humans, use the motion of counterparts to produce coordinated, social behaviors. But how do we resolve the discrete measures of communication and information theory with the continuous laws of motion and mechanics? Answering these questions is critical to developing expressive robotic systems that integrate seamlessly with natural counterparts a goal that has increasing urgency as robots move out of factories and into workplaces and homes. This talk presents this problem in an information-theoretic model (where artificial systems are modeled as a source communicating across a channel to a human receiver) and highlights how this model guides work in the Robotics, Automation, and Dance (RAD) Lab. The talk will present work in generating variable bipedal gait via parameters embedded as constraints in an optimization, predicting perceived affect when human viewers observe artificial gait across multiple environments, and imitating human motion on multiple robotic platforms. In addition to traditional tools in dynamics, control, and empirical measurement, these projects leverage qualitative observation, embodied movement practice, and artistic creation. Thus, the talk will also highlight how dancing with robots is critical to developing automation that functions correctly in human-built spaces.
CA, email@example.com, 5104953505