Manipulation and reasoning methods for embodied object search
Abstract/Contents
- Abstract
- Embodied agents often need to find objects to achieve downstream tasks, which makes it valuable to study solutions to the challenges posed by embodied object search. These challenges will depend on the environment the agent needs to search in: in small environments such as a bin or a shelf the primary challenges will relate to manipulation in clutter and perception in the presence occlusions, whereas in larger environments such as a room or an apartment the agent will also have to reason about where objects are likely to be and navigate to these locations. In this talk, I will present new methods for solving this variety of challenges -- with a focus on manipulation and reasoning -- in a variety of environments, and with a variety of techniques. Specifically, I will present formulations of the problem in the context of seeking to extract an object from a bin, to reveal a hidden object on a tabletop, to predict an object's location in a house, and to remember patterns of object movement in a variety of households. For each formulation, I will present a novel learning-based approach that expands on or pushes beyond what was achieved previously. These approaches will involve object segmentation, object recognition, grasp planning, teacher-aided reinforcement learning, procedural environment generation, graph neural networks, and more. I will conclude by discussing how these methods can be refined and combined to enable embodied agents to find objects in novel real world environments.
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 | 2023; ©2023 |
| Publication date | 2023; 2023 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Kurenkov, Andrey |
|---|---|
| Degree supervisor | Bohg, Jeannette, 1981- |
| Thesis advisor | Bohg, Jeannette, 1981- |
| Thesis advisor | Finn, Chelsea |
| Degree committee member | Finn, Chelsea |
| Associated with | Stanford University, School of Engineering |
| Associated with | Stanford University, Computer Science Department |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Andrey Kurenkov. |
|---|---|
| Note | Submitted to the Computer Science Department. |
| Thesis | Thesis Ph.D. Stanford University 2023. |
| Location | https://purl.stanford.edu/pk887jt7161 |
Access conditions
- Copyright
- © 2023 by Andrey Kurenkov
- 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...