Building robot intelligence by scaling human supervision
Abstract/Contents
- Abstract
- Large-scale human supervision has been at the heart of some of the most significant recent advances in domains such as computer vision and natural language processing, enabling near-human or even super-human performance on decades-old problems such as image recognition and question answering. However, robotics has not witnessed such success - the manipulation capabilities of today's robots pale in comparison to the wide range of tasks that we perform effortlessly on a daily basis. Developing systems and algorithms to collect and learn from large-scale human supervision could help bridge this gap in robot and human abilities. In this dissertation, I discuss my work, which aims to make human supervision a viable path towards developing intelligent and capable robots. I first discuss RoboTurk, a system we developed to collect large datasets filled with rich interactions that embody human-like manipulation abilities. Next, I discuss how robots can make use of these rich datasets to learn physical manipulation skills such as picking, placing, inserting, and assembling various objects. Together, these form a general paradigm for building capable robots through the use of large human datasets. Finally, I discuss how this paradigm can enable us to tackle a wider range of problem settings by collecting and leveraging these datasets in new ways.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2021; ©2021 |
Publication date | 2021; 2021 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Mandlekar, Ajay Uday | |
---|---|---|
Degree supervisor | Li, Fei Fei, 1976- | |
Degree supervisor | Savarese, Silvio | |
Thesis advisor | Li, Fei Fei, 1976- | |
Thesis advisor | Savarese, Silvio | |
Thesis advisor | Finn, Chelsea | |
Thesis advisor | Sadigh, Dorsa | |
Degree committee member | Finn, Chelsea | |
Degree committee member | Sadigh, Dorsa | |
Associated with | Stanford University, Department of Electrical Engineering |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Ajay Mandlekar. |
---|---|
Note | Submitted to the Department of Electrical Engineering. |
Thesis | Thesis Ph.D. Stanford University 2021. |
Location | https://purl.stanford.edu/pk658rz1153 |
Access conditions
- Copyright
- © 2021 by Ajay Uday Mandlekar
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...