Year: 2020
Type: dissertation
Abstract: Mobile manipulators consist of a mobile platform equipped with one or more robot arms and are of interest for a wide array of challenging tasks because of their extended workspace and dexterity. Typically, mobile manipulators are deployed in slow-motion collaborative robot scenarios, but they are capable of more challenging tasks. In this thesis, we consider scenarios where accurate high-speed motions are required. We proposed two different frameworks for this regime of tasks: (i) a modular framework that includes a bi-level motion planner and a learning-based controller; and (ii) an end-to-end framework that is based on deep reinforcement learning and simulation-to-reality transfer techniques. We evaluate our frameworks with a mobile manipulator platform through numerous high-speed ball catching experiments, where we show a success rate of 85.33% for real robots and 85.60% for simulated robots.
Source: TSpace (University of Toronto)
Author: Ke Dong
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3159446125