AI for Long-Term Autonomy
A PROPOSAL FOR A RA-L SPECIAL ISSUE ON:
AI FOR LONG-TERM AUTONOMY
T. DUCKETT, L. KUNZE, M. HANHEIDE, G. SIBLEY, AND
N. HAWES
TOPICAL AREA
Autonomous systems have a long history in the fields of Artificial Intelligence (AI) and Robotics. However,
only through recent advances in technology has it been possible to create autonomous systems capable of
operating in long-term, real-world scenarios. Examples include autonomous robots that operate outdoors
on land, air, water, and space, and indoors in offices, care homes, and factories. Designing, developing, and
maintaining intelligent autonomous systems that operate in real-world environments over long periods of
time, i.e. weeks, months, or years, poses many challenges. The proposed Special Issue will focus on such
challenges and on ways to overcome them using methods from AI.
The size of this community is estimated as several hundred researchers. The proposal follows the successful
organization of the ICRA 2016 Workshop on AI for Long-Term Autonomy, and previous events held at
major robotics conferences (e.g. ICRA Workshops on Long-term Autonomy, 2011-2014) and events on AI
and Robotics (http://ai-robotics.wikispaces.com/events).
MOTIVATION AND ENDORSEMENTS
The proposal of this Special Issue is motivated by the success of the Workshop “AI for Long-Term
Autonomy”, organized at the ICRA’16 conference on 16 May 2016
(https://sites.google.com/site/icra2016ailta/). The quality of the speakers and number of attendees (ca 200)
were very encouraging. In the absence of published proceedings of the Workshop, the important
contributions presented there could be missed by a wider public. The proposed Special Issue would also
encourage participation by other attendees of the workshop and the wider scientific community working on
related issues in AI for long-running robotic systems.
The ICRA’16 Workshop on AI for Long-Term Autonomy was endorsed by the IEEE Technical Committee on
Cognitive Robotics (please see endorsement letter in attachment).
OTHER RECENT SPECIAL ISSUES
To the best of our knowledge, there has never been a dedicated journal Special Issue on AI for Long-Term
Autonomy. The most closely related Special Issues include the following:
● Tim Barfoot , Jonathan Kelly and Gabe Sibley (editors). International Journal of Robotics Research.
Special Issue on Long-Term Autonomy . January 2014.
o This Special Issue covered two main themes: (i) localization and mapping for dynamic
environments, and (ii) lifelong learning for adaptive behavior. Highly relevant, although
AI methods here are limited to topics (i) and (ii). The proposed Special Issue will
potentially cover a wider range of AI methods for enabling long-term autonomy. The
field of long-term autonomy has also seen substantial progress in the past few years
with e.g. run times increasing from timescales of days to months, so the proposed SI
will further help to document the recent developments.
● Francesco Maurelli , Marc Carreras , Kanna Rajan ,and David Lane (editors). Autonomous Robots.
Special Issue on Long-Term Autonomy in Marine Robotics . Vol. 40, Issue 7, October 2016.
o Clearly relevant but specific to marine robotics.
● Paul Furgale, Cédric Pradalier, Timothy D. Barfoot (editors). Journal of Field Robotics. Special Issue
on Calibration for Field Robotics , Vol. 32, Issue 5, 2015.
o The editorial describes “questions of long-term autonomy” as motivation for the
Special Issue, where “calibration parameters will have to be determined efficiently and
maintained over the lifetime of a robot”. However, none of the selected papers actually
include long-term experiments.
● Kanna Rajan and Alessandro Saffiotti. Artificial Intelligence.
Special Issue on AI and Robotics . Volume 247, Pages 1-440 (June 2017)
o This SI covers research of integrated systems that use methods of AI in the context of
robotics. The proposed SI is complementary as it will provide a focus on systems that
use AI methods for enabling long-term autonomy.
● Robotics and Autonomous Systems: no related Special Issues found.
● IEEE Transactions on Robotics: no related Special Issues found.
CANDIDATE PAPERS
Below is a list of proposed papers for the Special Issue. Please see titles and abstracts of all papers as
attachment.
1) Towards Wide Long-Term Autonomy: Learning to Switch Among Multiple Task Models ,
Anahita Mohseni-Kabir and Manuela Veloso
2) Uncertainty-aware motion planning on multi-modal gaussian mixture fields , Chittaranjan S
Swaminathan, Tomasz Kucner, Martin Magnusson, and Achim J Lilienthal
3) Adaptive CNNs for Online Learning and Long-Term Autonomy, Fabio Ramo et al.
4) Quality-aware, probabilistic flow mapping with spatial information , Tomasz Kucner et al.
5) Teach, repeat, learn and improve: An introspective system for life-long teach-and-repeat
visual navigation, Tom Krajnik et al.
6) Warped hypertime models for long-term localisation, navigation and exploration in
changing environments, Tom Krajnik et al.
7) Advances in Synoptic Observations in Marine Robotics, Tor Arne Johansen, Kanna Rajan, and
Joao Sousa
8) Long-Term Deployment of Mobile Manipulators: Introspection and Self-Calibration for
Reliable Operation , Jonathan Kelly et al.
9) Long Term Low Latency Localization for Autonomous Vehicles, Michael Milford et al.
2
10) Long Term Intelligent Perception for Autonomous Underground Mining Vehicles, Michael
Milford et al.
11) Rhythmic Representations: Learning Periodic Visual Patterns for Sub-Linear Endlessly Scalable
Place Recognition, Michael Milford et al.
EDITORIAL CONTRIBUTION
At the Special Issue deadline, the Guest Editors will submit a Survey article of regular length (6 to 8 pages)
containing a thorough review of recent literature on AI for Long-Term Autonomy. Should the Survey paper
not be accepted, the Guest Editors will provide a two-page Editorial including an introduction of the Special
Issue and a discussion of the contents of all papers appearing in the Issue.
TIMELINE
The proposed Special Issue editorial timeline, consistent with the RA-L policy of publishing in max. 6
months from submission.
1) November 2, 2017 : first contact with the RA-L Editorial Board with a proposal draft
2) November 16, 2017 : the proposal is accepted and a Supervising Editor is nominated. A submission
deadline is fixed for February 24, 2018
3) December 1, 2017: A call for papers is launched on RA-L webpage. Guest Editors provide
maximum visibility by distributing to general and specialized mailing lists
4) February 24, 2018: submissions to Special Issue closes.
5) Normal RA-L review process schedule follows: First Decision by May 20, 2018; Final Decision by
August 1, 2017; Special Issue Published by September 1, 2018 .
CONFERENCE OPTION
The Special Issue will be associated with the 2018 IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS), October 1-5, 2018, Madrid, Spain and accepted articles will be presented at
the conference. We have aligned the timeline with the submission deadline of the general deadline of
RA-L articles with conference option (however, if necessary adjustments can be made). We made the
authors of the invited candidate proposals aware that accepted papers will be presented at IROS.
Similarly, we will make this clear in the CFP to this Special Issue.
3
A PROPOSAL FOR A RA-L SPECIAL ISSUE ON:
AI FOR LONG-TERM AUTONOMY
T. DUCKETT, L. KUNZE, M. HANHEIDE, G. SIBLEY, AND
N. HAWES
TOPICAL AREA
Autonomous systems have a long history in the fields of Artificial Intelligence (AI) and Robotics. However,
only through recent advances in technology has it been possible to create autonomous systems capable of
operating in long-term, real-world scenarios. Examples include autonomous robots that operate outdoors
on land, air, water, and space, and indoors in offices, care homes, and factories. Designing, developing, and
maintaining intelligent autonomous systems that operate in real-world environments over long periods of
time, i.e. weeks, months, or years, poses many challenges. The proposed Special Issue will focus on such
challenges and on ways to overcome them using methods from AI.
The size of this community is estimated as several hundred researchers. The proposal follows the successful
organization of the ICRA 2016 Workshop on AI for Long-Term Autonomy, and previous events held at
major robotics conferences (e.g. ICRA Workshops on Long-term Autonomy, 2011-2014) and events on AI
and Robotics (http://ai-robotics.wikispaces.com/events).
MOTIVATION AND ENDORSEMENTS
The proposal of this Special Issue is motivated by the success of the Workshop “AI for Long-Term
Autonomy”, organized at the ICRA’16 conference on 16 May 2016
(https://sites.google.com/site/icra2016ailta/). The quality of the speakers and number of attendees (ca 200)
were very encouraging. In the absence of published proceedings of the Workshop, the important
contributions presented there could be missed by a wider public. The proposed Special Issue would also
encourage participation by other attendees of the workshop and the wider scientific community working on
related issues in AI for long-running robotic systems.
The ICRA’16 Workshop on AI for Long-Term Autonomy was endorsed by the IEEE Technical Committee on
Cognitive Robotics (please see endorsement letter in attachment).
OTHER RECENT SPECIAL ISSUES
To the best of our knowledge, there has never been a dedicated journal Special Issue on AI for Long-Term
Autonomy. The most closely related Special Issues include the following:
● Tim Barfoot , Jonathan Kelly and Gabe Sibley (editors). International Journal of Robotics Research.
Special Issue on Long-Term Autonomy . January 2014.
o This Special Issue covered two main themes: (i) localization and mapping for dynamic
environments, and (ii) lifelong learning for adaptive behavior. Highly relevant, although
AI methods here are limited to topics (i) and (ii). The proposed Special Issue will
potentially cover a wider range of AI methods for enabling long-term autonomy. The
field of long-term autonomy has also seen substantial progress in the past few years
with e.g. run times increasing from timescales of days to months, so the proposed SI
will further help to document the recent developments.
● Francesco Maurelli , Marc Carreras , Kanna Rajan ,and David Lane (editors). Autonomous Robots.
Special Issue on Long-Term Autonomy in Marine Robotics . Vol. 40, Issue 7, October 2016.
o Clearly relevant but specific to marine robotics.
● Paul Furgale, Cédric Pradalier, Timothy D. Barfoot (editors). Journal of Field Robotics. Special Issue
on Calibration for Field Robotics , Vol. 32, Issue 5, 2015.
o The editorial describes “questions of long-term autonomy” as motivation for the
Special Issue, where “calibration parameters will have to be determined efficiently and
maintained over the lifetime of a robot”. However, none of the selected papers actually
include long-term experiments.
● Kanna Rajan and Alessandro Saffiotti. Artificial Intelligence.
Special Issue on AI and Robotics . Volume 247, Pages 1-440 (June 2017)
o This SI covers research of integrated systems that use methods of AI in the context of
robotics. The proposed SI is complementary as it will provide a focus on systems that
use AI methods for enabling long-term autonomy.
● Robotics and Autonomous Systems: no related Special Issues found.
● IEEE Transactions on Robotics: no related Special Issues found.
CANDIDATE PAPERS
Below is a list of proposed papers for the Special Issue. Please see titles and abstracts of all papers as
attachment.
1) Towards Wide Long-Term Autonomy: Learning to Switch Among Multiple Task Models ,
Anahita Mohseni-Kabir and Manuela Veloso
2) Uncertainty-aware motion planning on multi-modal gaussian mixture fields , Chittaranjan S
Swaminathan, Tomasz Kucner, Martin Magnusson, and Achim J Lilienthal
3) Adaptive CNNs for Online Learning and Long-Term Autonomy, Fabio Ramo et al.
4) Quality-aware, probabilistic flow mapping with spatial information , Tomasz Kucner et al.
5) Teach, repeat, learn and improve: An introspective system for life-long teach-and-repeat
visual navigation, Tom Krajnik et al.
6) Warped hypertime models for long-term localisation, navigation and exploration in
changing environments, Tom Krajnik et al.
7) Advances in Synoptic Observations in Marine Robotics, Tor Arne Johansen, Kanna Rajan, and
Joao Sousa
8) Long-Term Deployment of Mobile Manipulators: Introspection and Self-Calibration for
Reliable Operation , Jonathan Kelly et al.
9) Long Term Low Latency Localization for Autonomous Vehicles, Michael Milford et al.
2
10) Long Term Intelligent Perception for Autonomous Underground Mining Vehicles, Michael
Milford et al.
11) Rhythmic Representations: Learning Periodic Visual Patterns for Sub-Linear Endlessly Scalable
Place Recognition, Michael Milford et al.
EDITORIAL CONTRIBUTION
At the Special Issue deadline, the Guest Editors will submit a Survey article of regular length (6 to 8 pages)
containing a thorough review of recent literature on AI for Long-Term Autonomy. Should the Survey paper
not be accepted, the Guest Editors will provide a two-page Editorial including an introduction of the Special
Issue and a discussion of the contents of all papers appearing in the Issue.
TIMELINE
The proposed Special Issue editorial timeline, consistent with the RA-L policy of publishing in max. 6
months from submission.
1) November 2, 2017 : first contact with the RA-L Editorial Board with a proposal draft
2) November 16, 2017 : the proposal is accepted and a Supervising Editor is nominated. A submission
deadline is fixed for February 24, 2018
3) December 1, 2017: A call for papers is launched on RA-L webpage. Guest Editors provide
maximum visibility by distributing to general and specialized mailing lists
4) February 24, 2018: submissions to Special Issue closes.
5) Normal RA-L review process schedule follows: First Decision by May 20, 2018; Final Decision by
August 1, 2017; Special Issue Published by September 1, 2018 .
CONFERENCE OPTION
The Special Issue will be associated with the 2018 IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS), October 1-5, 2018, Madrid, Spain and accepted articles will be presented at
the conference. We have aligned the timeline with the submission deadline of the general deadline of
RA-L articles with conference option (however, if necessary adjustments can be made). We made the
authors of the invited candidate proposals aware that accepted papers will be presented at IROS.
Similarly, we will make this clear in the CFP to this Special Issue.
3
A Special Issue proposal by
F. Ficuciello, P. Falco and S. Calinon


The special issue presents a hot topic in robotic manipulation that is how to merge learning and control strategies to provide autonomy to robotic systems. This is a fundamental problem for robotic manipulation and it touches different functional system levels based on both visual and tactile perception and also affects the mechanical design of the robot parts that move and interact with the environment. The anthropomorphism contemplated in the mechanical design encourages the inspiration of learning techniques and control from the functioning of the human being. This implies the involvement of a large research community not only in the field of robotics but also the community of neuroscientists. The problem is highly multi-disciplinary, and different approaches from control theory, machine learning, and artificial intelligence, will be compared and possibly combined. 

Motivation

The topic of this Special Issue was the subject of a full-day workshop “Learning and control for autonomous manipulation systems: the role of dimensionality reduction”, organized at ICRA 2017. The large attendance (among 80-100 participants), the quality of the speakers, and the relevance of the addressed topics of research have been key elements for the success of the workshop. The Special Issue wants to collect an even broader scope of contributions, and make them available in archival format.

Topics of Interest

  • Dimensionality reduction in anthropomorphic design
  • Mechanical and motor synergies
  • Perceptual and sensorimotor synergies
  • The role of dimensionality reduction in learning strategies
  • The role of dimensionality reduction in control strategies

ENDORSEMENT

IEEE-RAS Technical Committee on Robotic Hands, Grasping, and Manipulation

TIMELINE

Special Issue Call Publication: August 1, 2017
Special Issue Submission Opens: October 5, 2017
Special Issue Submission Closes: October 15, 2017
First Decision Communicated to Authors: January 9, 2018
Final Decision Communicated to Authors: March 15, 2018
Accepted RAL Papers appear on Xplore: April 3, 2018

Easy Links