MORE | Spring 2023, Needs Review
Improving Behavior-Based Swarm Control Using Partially Specified Human Preferences
In this work, we explore a novel approach to improving behavior-based swarm control using partially specified human preferences. In order to address concerns about human-robot safety, we plan to incorporate human feedback in the form of user preferences over short clips of swarm behavior from non-expert users. To provide users with better expressiveness while specifying their preferences, these preferences are acquired on the basis of certain parameters that describe the motion of the swarm. These parameters are interdependent, and in order to learn the ideal trajectory of these parameters, we use Gaussian Processes to model the relationships between these parameters.
Robotics and autonomous systems
Hometown: Delhi, Delhi, India
Graduation date: Spring 2023