Page 135 |
Save page Remove page | Previous | 135 of 151 | Next |
|
small (250x250 max)
medium (500x500 max)
Large (1000x1000 max)
Extra Large
large ( > 500x500)
Full Resolution
All (PDF)
|
This page
All
|
priors into the meta-learned losses. We demonstrate wide applicability of this approach on a family
of control and classical machine learning problems.
6.1 Conclusion
There are a few key insights from our work. First, there is a lot of value in leveraging alternative
sensory modalities and robot collaborations when it comes to feedback scarcity. If a single
modality can not provide the full information about the state, other modalities, given a set of simple
spatio-temporal assumptions about the problem’s structure, can effectively complement the others.
Even for uniquely hard problems with a lot of uncertainty, effective solutions can be relatively
simple and involve a set of heuristics. Although heuristics often do not produce optimal solutions,
they retain better generalization capability since they do not rely on implicitly build models of
environments that can overfit to the training state distribution and can be fragile in the presence
of uncertainty. This is also seen from examples of human behavior. We tend to extract a set of
simple heuristics [Gigerenzer and Brighton, 2009] from our experiences instead of building precise
internal models of processes.
Simulators can provide an important prior for learning transferable policies using reinforcement
learning. We demonstrate in our work that even in well-studied problems RL can find uniquely
good solutions. We believe this comes from i) relaxation of assumptions in the hierarchical nature
of controllers that are usually required for provably stable solutions; and ii) RL does not always
find fragile controllers that rely on precise modelling, but can learn robust heuristics that generalize
beyond the assumptions of the simulator.
120
Object Description
| Title | Data scarcity in robotics: leveraging structural priors and representation learning |
| Author | Molchanov, Artem |
| Author email | a.molchanov86@gmail.com;molchano@usc.edu |
| Degree | Doctor of Philosophy |
| Document type | Dissertation |
| Degree program | Computer Science |
| School | Viterbi School of Engineering |
| Date defended/completed | 2020-05-11 |
| Date submitted | 2020-08-11 |
| Date approved | 2020-08-11 |
| Restricted until | 2020-08-11 |
| Date published | 2020-08-11 |
| Advisor (committee chair) | Sukhatme, Gaurav Suhas |
| Advisor (committee member) |
Ayanian, Nora Culbertson, Heather Gupta, Satyandra K. |
| Abstract | Recent advances in Artificial Intelligence have benefited significantly from access to large pools of data accompanied in many cases by labels, ground truth values, or perfect demonstrations. In robotics, however, such data are scarce or absent completely. Overcoming this issue is a major barrier to move robots from structured laboratory settings to the unstructured real world. In this dissertation, by leveraging structural priors and representation learning, we provide several solutions when data required to operate robotics systems is scarce or absent. ❧ In the first part of this dissertation we study sensory feedback scarcity. We show how to use high-dimensional alternative sensory modalities to extract data when primary sensory sources are absent. In a robot grasping setting, we address the problem of contact localization and solve it using multi-modal tactile feedback as the alternative source of information. We leverage multiple tactile modalities provided by electrodes and hydro-acoustic sensors to structure the problem as spatio-temporal inference. We employ the representational power of neural networks to acquire the complex mapping between tactile sensors and the contact locations. We also investigate scarce feedback due to the high cost of measurements. We study this problem in a challenging field robotics setting where multiple severely underactuated aquatic vehicles need to be coordinated. We show how to leverage collaboration among the vehicles and the spatio-temporal smoothness of the ocean currents as a prior to densify feedback about ocean currents in order to acquire better controllability. ❧ In the second part of this dissertation, we investigate scarcity of the data related to the desired task. We develop a method to efficiently leverage simulated dynamics priors to perform sim-to-real transfer of a control policy when no data about the target system is available. We investigate this problem in the scenario of sim-to-real transfer of low-level stabilizing quadrotor control policies. We demonstrate that we can learn robust policies in simulation and transfer them to the real system while acquiring no samples from the real quadrotor. Finally, we consider the general problem of learning a model with a very limited number of samples using meta-learned losses. We show how such losses can encode a prior structure about families of tasks to create well-behaved loss landscapes for efficient model optimization. We demonstrate the efficiency of our approach for learning policies and dynamics models in multiple robotics settings. |
| Keyword | robotics; machine learning; artificial intelligence |
| Language | English |
| Part of collection | University of Southern California dissertations and theses |
| Publisher (of the original version) | University of Southern California |
| Place of publication (of the original version) | Los Angeles, California |
| Publisher (of the digital version) | University of Southern California. Libraries |
| Provenance | Electronically uploaded by the author |
| Type | texts |
| Legacy record ID | usctheses-m |
| Contributing entity | University of Southern California |
| Rights | Molchanov, Artem |
| Physical access | The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given. |
| Repository name | University of Southern California Digital Library |
| Repository address | USC Digital Library, University of Southern California, University Park Campus MC 7002, 106 University Village, Los Angeles, California 90089-7002, USA |
| Repository email | cisadmin@lib.usc.edu |
| Filename | etd-MolchanovA-8923.pdf |
| Archival file | Volume13/etd-MolchanovA-8923.pdf |
Description
| Title | Page 135 |
| Full text | priors into the meta-learned losses. We demonstrate wide applicability of this approach on a family of control and classical machine learning problems. 6.1 Conclusion There are a few key insights from our work. First, there is a lot of value in leveraging alternative sensory modalities and robot collaborations when it comes to feedback scarcity. If a single modality can not provide the full information about the state, other modalities, given a set of simple spatio-temporal assumptions about the problem’s structure, can effectively complement the others. Even for uniquely hard problems with a lot of uncertainty, effective solutions can be relatively simple and involve a set of heuristics. Although heuristics often do not produce optimal solutions, they retain better generalization capability since they do not rely on implicitly build models of environments that can overfit to the training state distribution and can be fragile in the presence of uncertainty. This is also seen from examples of human behavior. We tend to extract a set of simple heuristics [Gigerenzer and Brighton, 2009] from our experiences instead of building precise internal models of processes. Simulators can provide an important prior for learning transferable policies using reinforcement learning. We demonstrate in our work that even in well-studied problems RL can find uniquely good solutions. We believe this comes from i) relaxation of assumptions in the hierarchical nature of controllers that are usually required for provably stable solutions; and ii) RL does not always find fragile controllers that rely on precise modelling, but can learn robust heuristics that generalize beyond the assumptions of the simulator. 120 |
