[robotics-worldwide] [meetings] Reminder: CFP RSS 2014 Workshop on "Learning Plans with Context From Human Signals" [Learning+Planning+HRI]
ashesh at cs.cornell.edu
Thu May 22 08:28:08 PDT 2014
Call for Papers
RSS Workshop on Learning Plans with Context From Human Signals
July 13, 2014 @ Berkeley, California, USA (full day)
In conjunction with the Robotics Science and Systems (RSS) 2014.
***This workshop aims at a broader audience and will bring together people
from machine learning, planning and HRI communities.
WORKSHOP DESCRIPTION / CALL FOR CONTRIBUTIONS:
Human environments such as homes, warehouses and offices are
very rich with the context of the objects and humans present and
the task to be performed. Robots should incorporate this rich context
and plan human-preferred motions. Furthermore, the robots should learn from
different kinds of human feedback, which can range from optimal
demonstrations to sub-optimal incremental signals. From an HRI perspective,
signals come in various forms and we need new machine learning techniques
to use these signals for meaningful motion planning. Through this workshop
we bring together people from the areas of machine learning, planning
and HRI to discuss how robots can learn to plan and act in context-rich
TOPICS of interest include:
* Importance of context in path planning.
* Understanding contextual information to plan desirable paths/trajectories.
* Different kind of user preferences and the form of context relevant to it.
* Learning from user feedback.
* Users can provide informative signals in many forms, such as optimal
demonstrations or sub-optimal feedback. We welcome different perspectives
on the ease of eliciting the feedback, informativeness of the signals and
* Learning from Demonstration and Imitation learning.
* Beyond Imitation: Apprenticeship and self-improvement.
* Overcoming computational limitations by leveraging learning and human
* Human robot collaboration.
* Intent Inference.
* Planning in Semi-Autonomy: Plan how to optimally support a human using
* Planning in Competition: Can we infer what the human is doing and use the
human as an action?
* Learning with large scale user data, and its advantages in generalizing
to new situations.
Pieter Abbeel, University of California Berkeley
Alan Fern, Oregon State University
Manuel Lopes, INRIA
Maja Mataric, University of Southern California
Julie Shah, MIT
Andrea Thomaz, Georgia Institute of Technology
Electronic submissions can be in the form of both extended abstracts
(2-3 pages) and full papers (up to 8 pages). They can be either a
presentation of work in progress or a summary of recent research
advances. Video supplemental materials are encouraged. Selected
contributions will be presented at the workshop as talks, spotlights
and/or posters. Accepted papers will not be archived.
Papers, abstracts and supplemental materials can be submitted at:
Note that the reviews will not be double blind -- please include the
authors and affiliations in your submission. And we are allowing
submissions up to 8 pages in length.
The submissions should be in the standard RSS format. Instructions and
templates (Latex and Word) can be found on the RSS 2014 website:
We encourage the submissions of live demos and working systems. A demo
submission should be in the format of an extended abstract with
recorded videos, under the DEMO category, separate from the main paper
if any. Given sufficient interests, a special demo session will be
organized at the workshop.
* Deadline for submissions: June 10th, 2014
* Notification of acceptance: June 25th, 2014
* Workshop: July 13th, 2014
Drew Bagnell, CMU
Ashesh Jain, Cornell University
Jan Peters, TU Darmstadt
Ashutosh Saxena, Cornell University
Department of Computer Science
More information about the robotics-worldwide