Wed, Jul 25, 2018 @ 02:00 PM - 04:00 PM
Title: From Active to Interactive 3D Object Recognition
Ph.D. Candidate: Bharath Sankaran
Date and Time: Wednesday, July 25th, 2Pm
Committee: Nora Ayanian (Chair), Gaurav Sukhatme, Geoffrey Spedding
If robots are to successfully migrate from controlled industrial settings to unstructured human environments, they should be capable of robust perception in densely cluttered, noisy environments, plagued with poor lighting. This will require robots to move towards building semantic representations of the environment by detecting objects of interest and accurately estimating their pose. In this thesis, I explore the use of movement and interaction to solve one of the fundamental problems of computer vision, object detection and pose estimation. I exploit the notion that perception is a process that is both active and exploratory and reformulate the problem of 3D object recognition as one of movement and interaction. I formalize these problems as information acquisition optimal control problems. I will present both myopic and non-myopic solutions to the information acquisition problem for 3D object recognition using tools from optimal control and inverse optimal control. I will also introduce efficient approximations to these optimal control problems by exploiting the nature of the information measure or exploiting the behavior of the static recognition system.
Audiences: Everyone Is Invited
Contact: Lizsl De Leon