---
_id: '17898'
abstract:
- lang: eng
  text: 'There is an ever-growing zoo of modern neural network models that can efficiently
    learn end-to-end control from visual observations. These advanced deep models,
    ranging from convolutional to Vision Transformers, from small to gigantic networks,
    have been extensively tested on offline image classification tasks. In this paper,
    we study these vision models with respect to the open-loop training to closed-loop
    generalization abilities, i.e., deployment realizes a causal feedback loop that
    is not present during training. This causality gap typically emerges in robotics
    applications such as autonomous driving, where a network is trained to imitate
    the control commands of a human. In this setting, two situations arise: 1) Closed-loop
    testing in-distribution, where the test environment shares properties with those
    of offline training data. 2) Closed-loop testing under distribution shifts and
    out-of-distribution. Contrary to recently reported results, we show that under
    proper training guidelines, all vision architectures perform indistinguishably
    well on in-distribution deployment, resolving the causality gap. In situation
    2, We observe that scale is the strongest factor in improving closed-loop generalization
    regardless of the choice of the model architecture. Our results predict the trend
    that in the future we will see larger and larger models being used in offline-training-online-deployment
    imitation learning tasks in robotic applications.'
acknowledgement: This work was partially supported in parts by the ERC-2020-AdG 101020093.
  Additionally, it was partially sponsored by the United States Air Force Research
  Laboratory and the United States Air Force Artificial Intelligence Accelerator and
  was accomplished under Cooperative Agreement Number FA8750-19-2-1000. The views
  and conclusions contained in this document are those of the authors and should not
  be interpreted as representing the official policies, either expressed or implied,
  of the United States Air Force or the U.S. Government. The U.S. Government is authorized
  to reproduce and distribute reprints for Government purposes notwithstanding any
  copyright notation herein. This work was further supported by The Boeing Company
  and the Office of Naval Research (ONR) Grant N00014-18-1-2830.
article_processing_charge: No
author:
- first_name: Mathias
  full_name: Lechner, Mathias
  id: 3DC22916-F248-11E8-B48F-1D18A9856A87
  last_name: Lechner
- first_name: Ramin
  full_name: Hasani, Ramin
  last_name: Hasani
- first_name: Alexander
  full_name: Amini, Alexander
  last_name: Amini
- first_name: Tsun Hsuan
  full_name: Wang, Tsun Hsuan
  last_name: Wang
- first_name: Thomas A
  full_name: Henzinger, Thomas A
  id: 40876CD8-F248-11E8-B48F-1D18A9856A87
  last_name: Henzinger
  orcid: 0000-0002-2985-7724
- first_name: Daniela
  full_name: Rus, Daniela
  last_name: Rus
citation:
  ama: 'Lechner M, Hasani R, Amini A, Wang TH, Henzinger TA, Rus D. Overparametrization
    helps offline-to-online generalization of closed-loop control from pixels. In:
    <i>Proceedings of the 2024 IEEE International Conference on Robotics and Automation</i>.
    IEEE; 2024:2774-2782. doi:<a href="https://doi.org/10.1109/ICRA57147.2024.10610284">10.1109/ICRA57147.2024.10610284</a>'
  apa: 'Lechner, M., Hasani, R., Amini, A., Wang, T. H., Henzinger, T. A., &#38; Rus,
    D. (2024). Overparametrization helps offline-to-online generalization of closed-loop
    control from pixels. In <i>Proceedings of the 2024 IEEE International Conference
    on Robotics and Automation</i> (pp. 2774–2782). Yokohama, Japan: IEEE. <a href="https://doi.org/10.1109/ICRA57147.2024.10610284">https://doi.org/10.1109/ICRA57147.2024.10610284</a>'
  chicago: Lechner, Mathias, Ramin Hasani, Alexander Amini, Tsun Hsuan Wang, Thomas
    A Henzinger, and Daniela Rus. “Overparametrization Helps Offline-to-Online Generalization
    of Closed-Loop Control from Pixels.” In <i>Proceedings of the 2024 IEEE International
    Conference on Robotics and Automation</i>, 2774–82. IEEE, 2024. <a href="https://doi.org/10.1109/ICRA57147.2024.10610284">https://doi.org/10.1109/ICRA57147.2024.10610284</a>.
  ieee: M. Lechner, R. Hasani, A. Amini, T. H. Wang, T. A. Henzinger, and D. Rus,
    “Overparametrization helps offline-to-online generalization of closed-loop control
    from pixels,” in <i>Proceedings of the 2024 IEEE International Conference on Robotics
    and Automation</i>, Yokohama, Japan, 2024, pp. 2774–2782.
  ista: 'Lechner M, Hasani R, Amini A, Wang TH, Henzinger TA, Rus D. 2024. Overparametrization
    helps offline-to-online generalization of closed-loop control from pixels. Proceedings
    of the 2024 IEEE International Conference on Robotics and Automation. ICRA: International
    Conference on Robotics and Automation, 2774–2782.'
  mla: Lechner, Mathias, et al. “Overparametrization Helps Offline-to-Online Generalization
    of Closed-Loop Control from Pixels.” <i>Proceedings of the 2024 IEEE International
    Conference on Robotics and Automation</i>, IEEE, 2024, pp. 2774–82, doi:<a href="https://doi.org/10.1109/ICRA57147.2024.10610284">10.1109/ICRA57147.2024.10610284</a>.
  short: M. Lechner, R. Hasani, A. Amini, T.H. Wang, T.A. Henzinger, D. Rus, in:,
    Proceedings of the 2024 IEEE International Conference on Robotics and Automation,
    IEEE, 2024, pp. 2774–2782.
conference:
  end_date: 2024-05-17
  location: Yokohama, Japan
  name: 'ICRA: International Conference on Robotics and Automation'
  start_date: 2024-05-13
date_created: 2024-09-08T22:01:13Z
date_published: 2024-08-08T00:00:00Z
date_updated: 2026-08-12T06:35:50Z
day: '08'
department:
- _id: ToHe
doi: 10.1109/ICRA57147.2024.10610284
ec_funded: 1
external_id:
  isi:
  - '001294576202044'
fulldoi: https://doi.org/10.1109/ICRA57147.2024.10610284
isi: 1
language:
- iso: eng
month: '08'
oa_version: None
page: 2774-2782
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
  call_identifier: H2020
  grant_number: '101020093'
  name: Vigilant Algorithmic Monitoring of Software
publication: Proceedings of the 2024 IEEE International Conference on Robotics and
  Automation
publication_identifier:
  isbn:
  - '9798350384574'
  issn:
  - 1050-4729
publication_status: published
publisher: IEEE
quality_controlled: '1'
scopus_import: '1'
status: public
title: Overparametrization helps offline-to-online generalization of closed-loop control
  from pixels
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2024'
...
---
_id: '12976'
abstract:
- lang: eng
  text: "3D printing based on continuous deposition of materials, such as filament-based
    3D printing, has seen widespread adoption thanks to its versatility in working
    with a wide range of materials. An important shortcoming of this type of technology
    is its limited multi-material capabilities. While there are simple hardware designs
    that enable multi-material printing in principle, the required software is heavily
    underdeveloped. A typical hardware design fuses together individual materials
    fed into a single chamber from multiple inlets before they are deposited. This
    design, however, introduces a time delay between the intended material mixture
    and its actual deposition. In this work, inspired by diverse path planning research
    in robotics, we show that this mechanical challenge can be addressed via improved
    printer control. We propose to formulate the search for optimal multi-material
    printing policies in a reinforcement\r\nlearning setup. We put forward a simple
    numerical deposition model that takes into account the non-linear material mixing
    and delayed material deposition. To validate our system we focus on color fabrication,
    a problem known for its strict requirements for varying material mixtures at a
    high spatial frequency. We demonstrate that our learned control policy outperforms
    state-of-the-art hand-crafted algorithms."
acknowledgement: This work is graciously supported by FWF Lise Meitner (Grant M 3319).
  Kang Liao sincerely thank Emiliano Luci, Chunyu Lin, and Yao Zhao for their huge
  support.
article_processing_charge: No
author:
- first_name: Kang
  full_name: Liao, Kang
  last_name: Liao
- first_name: Thibault
  full_name: Tricard, Thibault
  last_name: Tricard
- first_name: Michael
  full_name: Piovarci, Michael
  id: 62E473F4-5C99-11EA-A40E-AF823DDC885E
  last_name: Piovarci
  orcid: 0000-0002-5062-4474
- first_name: Hans-Peter
  full_name: Seidel, Hans-Peter
  last_name: Seidel
- first_name: Vahid
  full_name: Babaei, Vahid
  last_name: Babaei
citation:
  ama: 'Liao K, Tricard T, Piovarci M, Seidel H-P, Babaei V. Learning deposition policies
    for fused multi-material 3D printing. In: <i>2023 IEEE International Conference
    on Robotics and Automation</i>. Vol 2023. IEEE; 2023:12345-12352. doi:<a href="https://doi.org/10.1109/ICRA48891.2023.10160465">10.1109/ICRA48891.2023.10160465</a>'
  apa: 'Liao, K., Tricard, T., Piovarci, M., Seidel, H.-P., &#38; Babaei, V. (2023).
    Learning deposition policies for fused multi-material 3D printing. In <i>2023
    IEEE International Conference on Robotics and Automation</i> (Vol. 2023, pp. 12345–12352).
    London, United Kingdom: IEEE. <a href="https://doi.org/10.1109/ICRA48891.2023.10160465">https://doi.org/10.1109/ICRA48891.2023.10160465</a>'
  chicago: Liao, Kang, Thibault Tricard, Michael Piovarci, Hans-Peter Seidel, and
    Vahid Babaei. “Learning Deposition Policies for Fused Multi-Material 3D Printing.”
    In <i>2023 IEEE International Conference on Robotics and Automation</i>, 2023:12345–52.
    IEEE, 2023. <a href="https://doi.org/10.1109/ICRA48891.2023.10160465">https://doi.org/10.1109/ICRA48891.2023.10160465</a>.
  ieee: K. Liao, T. Tricard, M. Piovarci, H.-P. Seidel, and V. Babaei, “Learning deposition
    policies for fused multi-material 3D printing,” in <i>2023 IEEE International
    Conference on Robotics and Automation</i>, London, United Kingdom, 2023, vol.
    2023, pp. 12345–12352.
  ista: 'Liao K, Tricard T, Piovarci M, Seidel H-P, Babaei V. 2023. Learning deposition
    policies for fused multi-material 3D printing. 2023 IEEE International Conference
    on Robotics and Automation. ICRA: International Conference on Robotics and Automation
    vol. 2023, 12345–12352.'
  mla: Liao, Kang, et al. “Learning Deposition Policies for Fused Multi-Material 3D
    Printing.” <i>2023 IEEE International Conference on Robotics and Automation</i>,
    vol. 2023, IEEE, 2023, pp. 12345–52, doi:<a href="https://doi.org/10.1109/ICRA48891.2023.10160465">10.1109/ICRA48891.2023.10160465</a>.
  short: K. Liao, T. Tricard, M. Piovarci, H.-P. Seidel, V. Babaei, in:, 2023 IEEE
    International Conference on Robotics and Automation, IEEE, 2023, pp. 12345–12352.
conference:
  end_date: 2023-06-02
  location: London, United Kingdom
  name: 'ICRA: International Conference on Robotics and Automation'
  start_date: 2023-05-29
date_created: 2023-05-16T09:14:09Z
date_published: 2023-07-04T00:00:00Z
date_updated: 2025-04-15T07:43:52Z
day: '04'
ddc:
- '004'
department:
- _id: BeBi
doi: 10.1109/ICRA48891.2023.10160465
external_id:
  isi:
  - '001048371104068'
file:
- access_level: open_access
  checksum: daeaa67124777d88487f933ea3f77164
  content_type: application/pdf
  creator: mpiovarc
  date_created: 2023-05-16T09:12:05Z
  date_updated: 2023-05-16T09:12:05Z
  file_id: '12977'
  file_name: Liao2023.pdf
  file_size: 5367986
  relation: main_file
  success: 1
file_date_updated: 2023-05-16T09:12:05Z
fulldoi: https://doi.org/10.1109/ICRA48891.2023.10160465
has_accepted_license: '1'
intvolume: '      2023'
isi: 1
keyword:
- reinforcement learning
- deposition
- control
- color
- multi-filament
language:
- iso: eng
month: '07'
oa: 1
oa_version: Submitted Version
page: 12345-12352
project:
- _id: eb901961-77a9-11ec-83b8-f5c883a62027
  grant_number: M03319
  name: Perception-Aware Appearance Fabrication
publication: 2023 IEEE International Conference on Robotics and Automation
publication_identifier:
  eisbn:
  - '9798350323658'
  issn:
  - 1050-4729
publication_status: published
publisher: IEEE
quality_controlled: '1'
scopus_import: '1'
status: public
title: Learning deposition policies for fused multi-material 3D printing
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 2023
year: '2023'
...
---
_id: '12010'
abstract:
- lang: eng
  text: World models learn behaviors in a latent imagination space to enhance the
    sample-efficiency of deep reinforcement learning (RL) algorithms. While learning
    world models for high-dimensional observations (e.g., pixel inputs) has become
    practicable on standard RL benchmarks and some games, their effectiveness in real-world
    robotics applications has not been explored. In this paper, we investigate how
    such agents generalize to real-world autonomous vehicle control tasks, where advanced
    model-free deep RL algorithms fail. In particular, we set up a series of time-lap
    tasks for an F1TENTH racing robot, equipped with a high-dimensional LiDAR sensor,
    on a set of test tracks with a gradual increase in their complexity. In this continuous-control
    setting, we show that model-based agents capable of learning in imagination substantially
    outperform model-free agents with respect to performance, sample efficiency, successful
    task completion, and generalization. Moreover, we show that the generalization
    ability of model-based agents strongly depends on the choice of their observation
    model. We provide extensive empirical evidence for the effectiveness of world
    models provided with long enough memory horizons in sim2real tasks.
acknowledgement: L.B. was supported by the Doctoral College Resilient Embedded Systems.
  M.L. was supported in part by the ERC2020-AdG 101020093 and the Austrian Science
  Fund (FWF) under grant Z211-N23 (Wittgenstein Award). R.H. and D.R. were supported
  by The Boeing Company and the Office of Naval Research (ONR) Grant N00014-18-1-2830.
  R.G. was partially supported by the Horizon-2020 ECSEL Project grant No. 783163
  (iDev40) and A.B. by FFG Project ADEX.
article_processing_charge: No
arxiv: 1
author:
- first_name: Axel
  full_name: Brunnbauer, Axel
  last_name: Brunnbauer
- first_name: Luigi
  full_name: Berducci, Luigi
  last_name: Berducci
- first_name: Andreas
  full_name: Brandstatter, Andreas
  last_name: Brandstatter
- first_name: Mathias
  full_name: Lechner, Mathias
  id: 3DC22916-F248-11E8-B48F-1D18A9856A87
  last_name: Lechner
- first_name: Ramin
  full_name: Hasani, Ramin
  last_name: Hasani
- first_name: Daniela
  full_name: Rus, Daniela
  last_name: Rus
- first_name: Radu
  full_name: Grosu, Radu
  last_name: Grosu
citation:
  ama: 'Brunnbauer A, Berducci L, Brandstatter A, et al. Latent imagination facilitates
    zero-shot transfer in autonomous racing. In: <i>2022 International Conference
    on Robotics and Automation</i>. IEEE; 2022:7513-7520. doi:<a href="https://doi.org/10.1109/ICRA46639.2022.9811650">10.1109/ICRA46639.2022.9811650</a>'
  apa: 'Brunnbauer, A., Berducci, L., Brandstatter, A., Lechner, M., Hasani, R., Rus,
    D., &#38; Grosu, R. (2022). Latent imagination facilitates zero-shot transfer
    in autonomous racing. In <i>2022 International Conference on Robotics and Automation</i>
    (pp. 7513–7520). Philadelphia, PA, United States: IEEE. <a href="https://doi.org/10.1109/ICRA46639.2022.9811650">https://doi.org/10.1109/ICRA46639.2022.9811650</a>'
  chicago: Brunnbauer, Axel, Luigi Berducci, Andreas Brandstatter, Mathias Lechner,
    Ramin Hasani, Daniela Rus, and Radu Grosu. “Latent Imagination Facilitates Zero-Shot
    Transfer in Autonomous Racing.” In <i>2022 International Conference on Robotics
    and Automation</i>, 7513–20. IEEE, 2022. <a href="https://doi.org/10.1109/ICRA46639.2022.9811650">https://doi.org/10.1109/ICRA46639.2022.9811650</a>.
  ieee: A. Brunnbauer <i>et al.</i>, “Latent imagination facilitates zero-shot transfer
    in autonomous racing,” in <i>2022 International Conference on Robotics and Automation</i>,
    Philadelphia, PA, United States, 2022, pp. 7513–7520.
  ista: 'Brunnbauer A, Berducci L, Brandstatter A, Lechner M, Hasani R, Rus D, Grosu
    R. 2022. Latent imagination facilitates zero-shot transfer in autonomous racing.
    2022 International Conference on Robotics and Automation. ICRA: International
    Conference on Robotics and Automation, 7513–7520.'
  mla: Brunnbauer, Axel, et al. “Latent Imagination Facilitates Zero-Shot Transfer
    in Autonomous Racing.” <i>2022 International Conference on Robotics and Automation</i>,
    IEEE, 2022, pp. 7513–20, doi:<a href="https://doi.org/10.1109/ICRA46639.2022.9811650">10.1109/ICRA46639.2022.9811650</a>.
  short: A. Brunnbauer, L. Berducci, A. Brandstatter, M. Lechner, R. Hasani, D. Rus,
    R. Grosu, in:, 2022 International Conference on Robotics and Automation, IEEE,
    2022, pp. 7513–7520.
conference:
  end_date: 2022-05-27
  location: Philadelphia, PA, United States
  name: 'ICRA: International Conference on Robotics and Automation'
  start_date: 2022-05-23
date_created: 2022-09-04T22:02:02Z
date_published: 2022-07-12T00:00:00Z
date_updated: 2025-09-10T09:39:53Z
day: '12'
department:
- _id: ToHe
doi: 10.1109/ICRA46639.2022.9811650
ec_funded: 1
external_id:
  arxiv:
  - '2103.04909'
  isi:
  - '000941277600124'
fulldoi: https://doi.org/10.1109/ICRA46639.2022.9811650
isi: 1
language:
- iso: eng
main_file_link:
- open_access: '1'
  url: https://doi.org/10.48550/arXiv.2103.04909
month: '07'
oa: 1
oa_version: Preprint
page: 7513-7520
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
  call_identifier: H2020
  grant_number: '101020093'
  name: Vigilant Algorithmic Monitoring of Software
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
  call_identifier: FWF
  grant_number: Z211
  name: Formal methods for the design and analysis of complex systems
publication: 2022 International Conference on Robotics and Automation
publication_identifier:
  isbn:
  - '9781728196817'
  issn:
  - 1050-4729
publication_status: published
publisher: IEEE
quality_controlled: '1'
scopus_import: '1'
status: public
title: Latent imagination facilitates zero-shot transfer in autonomous racing
type: conference
user_id: 317138e5-6ab7-11ef-aa6d-ffef3953e345
year: '2022'
...
---
OA_place: repository
OA_type: green
_id: '10666'
abstract:
- lang: eng
  text: Adversarial training is an effective method to train deep learning models
    that are resilient to norm-bounded perturbations, with the cost of nominal performance
    drop. While adversarial training appears to enhance the robustness and safety
    of a deep model deployed in open-world decision-critical applications, counterintuitively,
    it induces undesired behaviors in robot learning settings. In this paper, we show
    theoretically and experimentally that neural controllers obtained via adversarial
    training are subjected to three types of defects, namely transient, systematic,
    and conditional errors. We first generalize adversarial training to a safety-domain
    optimization scheme allowing for more generic specifications. We then prove that
    such a learning process tends to cause certain error profiles. We support our
    theoretical results by a thorough experimental safety analysis in a robot-learning
    task. Our results suggest that adversarial training is not yet ready for robot
    learning.
acknowledgement: M.L. and T.A.H. are supported in part by the Austrian Science Fund
  (FWF) under grant Z211-N23 (Wittgenstein Award). R.H. and D.R. are supported by
  Boeing and R.G. by Horizon-2020 ECSEL Project grant no. 783163 (iDev40).
article_processing_charge: No
arxiv: 1
author:
- first_name: Mathias
  full_name: Lechner, Mathias
  id: 3DC22916-F248-11E8-B48F-1D18A9856A87
  last_name: Lechner
- first_name: Ramin
  full_name: Hasani, Ramin
  last_name: Hasani
- first_name: Radu
  full_name: Grosu, Radu
  last_name: Grosu
- first_name: Daniela
  full_name: Rus, Daniela
  last_name: Rus
- first_name: Thomas A
  full_name: Henzinger, Thomas A
  id: 40876CD8-F248-11E8-B48F-1D18A9856A87
  last_name: Henzinger
  orcid: 0000-0002-2985-7724
citation:
  ama: 'Lechner M, Hasani R, Grosu R, Rus D, Henzinger TA. Adversarial training is
    not ready for robot learning. In: <i>2021 IEEE International Conference on Robotics
    and Automation</i>. IEEE; 2021:4140-4147. doi:<a href="https://doi.org/10.1109/ICRA48506.2021.9561036">10.1109/ICRA48506.2021.9561036</a>'
  apa: 'Lechner, M., Hasani, R., Grosu, R., Rus, D., &#38; Henzinger, T. A. (2021).
    Adversarial training is not ready for robot learning. In <i>2021 IEEE International
    Conference on Robotics and Automation</i> (pp. 4140–4147). Xi’an, China: IEEE.
    <a href="https://doi.org/10.1109/ICRA48506.2021.9561036">https://doi.org/10.1109/ICRA48506.2021.9561036</a>'
  chicago: Lechner, Mathias, Ramin Hasani, Radu Grosu, Daniela Rus, and Thomas A Henzinger.
    “Adversarial Training Is Not Ready for Robot Learning.” In <i>2021 IEEE International
    Conference on Robotics and Automation</i>, 4140–47. IEEE, 2021. <a href="https://doi.org/10.1109/ICRA48506.2021.9561036">https://doi.org/10.1109/ICRA48506.2021.9561036</a>.
  ieee: M. Lechner, R. Hasani, R. Grosu, D. Rus, and T. A. Henzinger, “Adversarial
    training is not ready for robot learning,” in <i>2021 IEEE International Conference
    on Robotics and Automation</i>, Xi’an, China, 2021, pp. 4140–4147.
  ista: 'Lechner M, Hasani R, Grosu R, Rus D, Henzinger TA. 2021. Adversarial training
    is not ready for robot learning. 2021 IEEE International Conference on Robotics
    and Automation. ICRA: International Conference on Robotics and Automation, 4140–4147.'
  mla: Lechner, Mathias, et al. “Adversarial Training Is Not Ready for Robot Learning.”
    <i>2021 IEEE International Conference on Robotics and Automation</i>, IEEE, 2021,
    pp. 4140–47, doi:<a href="https://doi.org/10.1109/ICRA48506.2021.9561036">10.1109/ICRA48506.2021.9561036</a>.
  short: M. Lechner, R. Hasani, R. Grosu, D. Rus, T.A. Henzinger, in:, 2021 IEEE International
    Conference on Robotics and Automation, IEEE, 2021, pp. 4140–4147.
conference:
  end_date: 2021-06-05
  location: Xi'an, China
  name: 'ICRA: International Conference on Robotics and Automation'
  start_date: 2021-05-30
das_tickbox: '1'
date_created: 2022-01-25T15:44:54Z
date_published: 2021-06-01T00:00:00Z
date_updated: 2026-07-07T06:20:35Z
day: '01'
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
doi: 10.1109/ICRA48506.2021.9561036
external_id:
  arxiv:
  - '2103.08187'
  isi:
  - '000765738803040'
fulldoi: https://doi.org/10.1109/ICRA48506.2021.9561036
has_accepted_license: '1'
isi: 1
language:
- iso: eng
main_file_link:
- open_access: '1'
  url: https://arxiv.org/abs/2103.08187
month: '06'
oa: 1
oa_version: Preprint
page: 4140-4147
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
  call_identifier: FWF
  grant_number: Z211
  name: Formal methods for the design and analysis of complex systems
publication: 2021 IEEE International Conference on Robotics and Automation
publication_identifier:
  eisbn:
  - 978-1-7281-9077-8
  eissn:
  - 2577-087X
  isbn:
  - 978-1-7281-9078-5
  issn:
  - 1050-4729
publication_status: published
publisher: IEEE
quality_controlled: '1'
related_material:
  record:
  - id: '11362'
    relation: dissertation_contains
    status: public
scopus_import: '1'
status: public
title: Adversarial training is not ready for robot learning
tmp:
  image: /images/cc_by_nc_nd.png
  legal_code_url: https://creativecommons.org/licenses/by-nc-nd/3.0/legalcode
  name: Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported (CC BY-NC-ND
    3.0)
  short: CC BY-NC-ND (3.0)
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2021'
...
---
_id: '8704'
abstract:
- lang: eng
  text: Traditional robotic control suits require profound task-specific knowledge
    for designing, building and testing control software. The rise of Deep Learning
    has enabled end-to-end solutions to be learned entirely from data, requiring minimal
    knowledge about the application area. We design a learning scheme to train end-to-end
    linear dynamical systems (LDS)s by gradient descent in imitation learning robotic
    domains. We introduce a new regularization loss component together with a learning
    algorithm that improves the stability of the learned autonomous system, by forcing
    the eigenvalues of the internal state updates of an LDS to be negative reals.
    We evaluate our approach on a series of real-life and simulated robotic experiments,
    in comparison to linear and nonlinear Recurrent Neural Network (RNN) architectures.
    Our results show that our stabilizing method significantly improves test performance
    of LDS, enabling such linear models to match the performance of contemporary nonlinear
    RNN architectures. A video of the obstacle avoidance performance of our method
    on a mobile robot, in unseen environments, compared to other methods can be viewed
    at https://youtu.be/mhEsCoNao5E.
acknowledgement: M.L. is supported in parts by the Austrian Science Fund (FWF) under
  grant Z211-N23 (Wittgenstein Award). R.H., and R.G. are partially supported by the
  Horizon-2020 ECSELProject grant No. 783163 (iDev40), and the Austrian Research Promotion
  Agency (FFG), Project No. 860424. R.H. and D.R. is partially supported by the Boeing
  Company.
alternative_title:
- ICRA
article_processing_charge: No
author:
- first_name: Mathias
  full_name: Lechner, Mathias
  id: 3DC22916-F248-11E8-B48F-1D18A9856A87
  last_name: Lechner
- first_name: Ramin
  full_name: Hasani, Ramin
  last_name: Hasani
- first_name: Daniela
  full_name: Rus, Daniela
  last_name: Rus
- first_name: Radu
  full_name: Grosu, Radu
  last_name: Grosu
citation:
  ama: 'Lechner M, Hasani R, Rus D, Grosu R. Gershgorin loss stabilizes the recurrent
    neural network compartment of an end-to-end robot learning scheme. In: <i>Proceedings
    - IEEE International Conference on Robotics and Automation</i>. IEEE; 2020:5446-5452.
    doi:<a href="https://doi.org/10.1109/ICRA40945.2020.9196608">10.1109/ICRA40945.2020.9196608</a>'
  apa: 'Lechner, M., Hasani, R., Rus, D., &#38; Grosu, R. (2020). Gershgorin loss
    stabilizes the recurrent neural network compartment of an end-to-end robot learning
    scheme. In <i>Proceedings - IEEE International Conference on Robotics and Automation</i>
    (pp. 5446–5452). Paris, France: IEEE. <a href="https://doi.org/10.1109/ICRA40945.2020.9196608">https://doi.org/10.1109/ICRA40945.2020.9196608</a>'
  chicago: Lechner, Mathias, Ramin Hasani, Daniela Rus, and Radu Grosu. “Gershgorin
    Loss Stabilizes the Recurrent Neural Network Compartment of an End-to-End Robot
    Learning Scheme.” In <i>Proceedings - IEEE International Conference on Robotics
    and Automation</i>, 5446–52. IEEE, 2020. <a href="https://doi.org/10.1109/ICRA40945.2020.9196608">https://doi.org/10.1109/ICRA40945.2020.9196608</a>.
  ieee: M. Lechner, R. Hasani, D. Rus, and R. Grosu, “Gershgorin loss stabilizes the
    recurrent neural network compartment of an end-to-end robot learning scheme,”
    in <i>Proceedings - IEEE International Conference on Robotics and Automation</i>,
    Paris, France, 2020, pp. 5446–5452.
  ista: 'Lechner M, Hasani R, Rus D, Grosu R. 2020. Gershgorin loss stabilizes the
    recurrent neural network compartment of an end-to-end robot learning scheme. Proceedings
    - IEEE International Conference on Robotics and Automation. ICRA: International
    Conference on Robotics and Automation, ICRA, , 5446–5452.'
  mla: Lechner, Mathias, et al. “Gershgorin Loss Stabilizes the Recurrent Neural Network
    Compartment of an End-to-End Robot Learning Scheme.” <i>Proceedings - IEEE International
    Conference on Robotics and Automation</i>, IEEE, 2020, pp. 5446–52, doi:<a href="https://doi.org/10.1109/ICRA40945.2020.9196608">10.1109/ICRA40945.2020.9196608</a>.
  short: M. Lechner, R. Hasani, D. Rus, R. Grosu, in:, Proceedings - IEEE International
    Conference on Robotics and Automation, IEEE, 2020, pp. 5446–5452.
conference:
  end_date: 2020-08-31
  location: Paris, France
  name: 'ICRA: International Conference on Robotics and Automation'
  start_date: 2020-05-31
date_created: 2020-10-25T23:01:19Z
date_published: 2020-05-01T00:00:00Z
date_updated: 2025-07-10T12:01:21Z
day: '01'
ddc:
- '000'
department:
- _id: ToHe
doi: 10.1109/ICRA40945.2020.9196608
external_id:
  isi:
  - '000712319503110'
file:
- access_level: open_access
  checksum: fccf7b986ac78046918a298cc6849a50
  content_type: application/pdf
  creator: dernst
  date_created: 2020-11-06T10:58:49Z
  date_updated: 2020-11-06T10:58:49Z
  file_id: '8733'
  file_name: 2020_ICRA_Lechner.pdf
  file_size: 1070010
  relation: main_file
  success: 1
file_date_updated: 2020-11-06T10:58:49Z
fulldoi: https://doi.org/10.1109/ICRA40945.2020.9196608
has_accepted_license: '1'
isi: 1
language:
- iso: eng
month: '05'
oa: 1
oa_version: Submitted Version
page: 5446-5452
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
  call_identifier: FWF
  grant_number: Z211
  name: Formal methods for the design and analysis of complex systems
publication: Proceedings - IEEE International Conference on Robotics and Automation
publication_identifier:
  isbn:
  - '9781728173955'
  issn:
  - 1050-4729
publication_status: published
publisher: IEEE
quality_controlled: '1'
scopus_import: '1'
status: public
title: Gershgorin loss stabilizes the recurrent neural network compartment of an end-to-end
  robot learning scheme
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2020'
...