Page 105 |
Save page Remove page | Previous | 105 of 151 | Next |
|
small (250x250 max)
medium (500x500 max)
Large (1000x1000 max)
Extra Large
large ( > 500x500)
Full Resolution
All (PDF)
|
This page
All
|
these tests we do not perform comparison with the “Mellinger” controller since it could not recover
properly from deviations from its goal position larger than 0.5m. This controller is mainly tuned
for trajectory tracking with closely located points provided as full state vectors. Our policy, on the
other hand, shows a substantial level of robustness on all three platforms. It performs especially
well on the Crazyflie platform recovering from 80% of all throws and up to 100% throws with
moderate attitude changes (B 35°)¶. Interestingly, it can even recover in more than half of scenarios
after hitting the ground.
The policy also shows substantial level of robustness on other quadrotors. Similar to the
Crazyflie platform, the throws with moderate attitude change do not cause serious problems to any
of the platforms and they recover in C 90% of trials. Stronger attitude disturbances are significantly
harder and we observe roughly 50% recovery rate on average.
More mild tests, like light pushes and pulling from the hover state, do not cause failures.
One observation that we make is that all policies learned some preference on yaw orientation
that it tries to stabilize although we did not provide any yaw-related costs apart from the cost on
angular velocities. We hypothesize that the policy seeks a “home” yaw angle because it becomes
unnecessary to reason about symmetry of rotation in the xy plane if the quadrotor is always
oriented in the same direction.
Another surprising observation is that the control policies can deal with much higher initial
velocities than those encountered in training (B 1 m/s). In practice, initial velocities in our tests
often exceed 3 m/s (see Fig. 4.4 for an example of a recovery trajectory). The policies can take-off
from the ground, thus overcoming the near-ground airflow effects. They can also fly from distances
far exceeding the boundaries of the position initialization box observed in the training. All these
¶Computed as a 95-th percentile in our experiments.
90
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 105 |
| Full text | these tests we do not perform comparison with the “Mellinger” controller since it could not recover properly from deviations from its goal position larger than 0.5m. This controller is mainly tuned for trajectory tracking with closely located points provided as full state vectors. Our policy, on the other hand, shows a substantial level of robustness on all three platforms. It performs especially well on the Crazyflie platform recovering from 80% of all throws and up to 100% throws with moderate attitude changes (B 35°)¶. Interestingly, it can even recover in more than half of scenarios after hitting the ground. The policy also shows substantial level of robustness on other quadrotors. Similar to the Crazyflie platform, the throws with moderate attitude change do not cause serious problems to any of the platforms and they recover in C 90% of trials. Stronger attitude disturbances are significantly harder and we observe roughly 50% recovery rate on average. More mild tests, like light pushes and pulling from the hover state, do not cause failures. One observation that we make is that all policies learned some preference on yaw orientation that it tries to stabilize although we did not provide any yaw-related costs apart from the cost on angular velocities. We hypothesize that the policy seeks a “home” yaw angle because it becomes unnecessary to reason about symmetry of rotation in the xy plane if the quadrotor is always oriented in the same direction. Another surprising observation is that the control policies can deal with much higher initial velocities than those encountered in training (B 1 m/s). In practice, initial velocities in our tests often exceed 3 m/s (see Fig. 4.4 for an example of a recovery trajectory). The policies can take-off from the ground, thus overcoming the near-ground airflow effects. They can also fly from distances far exceeding the boundaries of the position initialization box observed in the training. All these ¶Computed as a 95-th percentile in our experiments. 90 |
