[robotics-worldwide] [meetings] CfP: IROS WS - Automating Robot Experiments

Oliver Kroemer oliverkro at googlemail.com
Tue Jul 17 22:03:53 PDT 2018


IROS 2018 Workshop on:
Automating Robot Experiments - Manipulation and Learning


Important Dates:
Early-Bird Submission deadline:  July 25th
Standard Submission deadline:   August 31st
Workshop Date:   October 5th

Modern robotics research requires massive amounts of data. New algorithms
are constantly being developed and the requirements for rigorous
benchmarking experiments are steadily increasing. Many of these algorithms
are themselves data-driven learning methods, which compounds the need for
both training and testing data. Acquiring large amounts of data is
unfortunately not a trivial task for many areas of robotics research,
especially when the data collection involves the robot interacting with its
environment. Manipulation experiments often require objects to be reset
between trials. Similarly, learning locomotion on deformable terrain may
require the ground to be reset between experiments. In practice, the
resetting of the environment is often performed manually by researchers,
which is a tedious and time-consuming process. The reliance on manual human
intervention is often a major bottleneck in the experiment pipeline and
tends to be the critical factor in limiting the amount of data that can be
reasonably acquired. Hence, to collect the required data and perform
rigorous evaluations, we will need to automate robot experiments. In this
workshop, we will bring together members of the robotics research community
to discuss different approaches for increasing the autonomy of robot
experiments. We will discuss techniques for controlling experiment
environments, performing simulated evaluations, and running experiments
autonomously for extended periods of time. The workshop will address the
following questions:

- What are the key challenges for automating robot experiments?
- What are common techniques for increasing the autonomy of the evaluation
- What safety factors need to be considered when running robot experiments
- What is the role of simulation in training and evaluating methods?
- What other approaches can be used to address the data limitations in
robotics research?

We are soliciting submissions of extended abstracts of 1-2 pages on recent
work that is relevant to the workshop. We especially encourage any work
that explores automated experiments, large robotics datasets for
training/testing, simulations for training/testing, safety in experiments,
benchmarking, or deep learning. Abstracts should be submitted as PDF files
using the RAS workshop format files found here:


Submissions should be submitted by email to:
oliverkro at googlemail.com

Submission emails should include "[AutoRobExpWS]" in the subject line.
Abstracts will be single-blind reviewed on the basis of relevance,
significance, and clarity. The accepted contributions will be presented as
posters during the workshop.

Early-bird Submission deadline: July 25th (we will aim for next-day reviews)
Standard Submission deadline:   August 31st

Oliver Kroemer (CMU), Lerrel Pinto (CMU), and Katharina Muelling (CMU)

More information about the robotics-worldwide mailing list