Dear ROS users,

I'd like to bring to your attention two internships currently open at PAL Robotics, which are expected to use ROS extensively. Details can be found in the following links:
- Robot skills: learning to use the right skill
- Becoming skilled: acquiring novel skills with imitation learning and motion planning

Start dates are flexible, and can be as early as January 2013. These two internships are the follow-up of a previous project that compared motion generation (implemented with sampling-based planners) and motion recall (implemented with dynamic motion primitives) for reaching tasks (video).

Best,

Adolfo Rodríguez Tsouroukdissian.

---------- Text description of opening follows ----------

Robot skills: learning to use the right skill.

In the near future, robotic products and services will become an integral part
of our daily life. One key capability of such robots is that they must have a set of skills which
are able to solve a wide variety of everyday tasks in human environments.

Currently, a popular approach is to implement such skills as Dynamic Movement Primitives, which have
several very desirable properties in terms of generalization, compactness, computational load, and
on-line obstacle avoidance. In a previous project, Dynamic Movement Primitives were successfully
implemented and executed on the REEM robot, leading to efficient, predictable movement.

Rather than considering individual skills, the goal of this project is to provide the REEM robot
with a whole set of skills for everyday manipulation, and to learn in which task contexts these
skills can be successfully executed. In particular, the project will consist of the following tasks:

1. Implementing a software infrastructure for representing, storing, and maintaining libraries of
skills, represented as Dynamic Movement Primitives. This software will be developed within the Robot
Operating System (ROS).

2. Acquiring a set of 5-10 skills for the REEM through imitation learning. This will involve the use
of the Kinect sensor to track human motion, for which software is already available.

3. Learning for which tasks a skill should/can be used. For instance, some reaching skills may only
work for tables of certain heights. Predicting skill execution success from task features (for
instance table height, position of the object on the table) will be done with supervised learning,
based on trial-and-error exploration.

4. Evaluating the generalization and robustness of this skill-based approach on the REEM robot in
the context of real-world tasks.

Good C++ programming skills, as well as familiarity with Linux user development tools are required. Familiarity with ROS is desirable, but not compulsory.

If you're interested, please submit your resume to: recruit@pal-robotics.com.

Becoming skilled: acquiring novel skills with imitation learning and motion planning.

In the near future, robotic products and services will become an integral part
of our daily life. One key capability of such robots is that they must have a set of skills which
are able to solve a wide variety of everyday tasks in human environments.

Currently, a popular approach is to implement such skills as Dynamic Movement Primitives, which have
several very desirable properties in terms of generalization, compactness, computational load, and
on-line obstacle avoidance. In a previous project, Dynamic Movement Primitives were successfully
implemented and executed on the REEM robot, leading to efficient, predictable movement.

In this project, libraries of skills (represented as Dynamic Movement Primitives) should be
automatically extended with novel skills. This is essential if none of the skills in the library is
able to solve the current task. Two methods for extending such libraries will be investigated:
imitation learning and initializing skills with motion plans generated by sampling-based motion
planners. The project will consist of the following tasks:

1. Implementing a software infrastructure for representing, storing, maintaining, and extending
libraries of skills, represented as Dynamic Movement Primitives. This software will be developed
within the Robot Operating System (ROS). This task will be conducted in unison with its sister
project: Robot skills: learning to use the right skill.

2. Automatically initializing novel Dynamic Movement Primitives with the output of sampling-based
motion planners.

3. Automatically initializing novel skills through imitation learning. This will involve the use of
the Kinect sensor to track human motion, for which software is already available.

4. Optimizing novel skills with stochastic optimization techniques.

5. Evaluating the autonomy and robustness of the approach on the REEM robot in the context of real-
world tasks.

Good C++ programming skills, as well as familiarity with Linux user development tools are required. Familiarity with ROS is desirable, but not compulsory.

If you're interested, please submit your resume to: recruit@pal-robotics.com.

--
Adolfo Rodríguez Tsouroukdissian
Senior robotics engineer
adolfo.rodriguez@pal-robotics.com
http://www.pal-robotics.com

PAL ROBOTICS S.L
c/ Pujades 77-79, 4º4ª
08005 Barcelona, Spain.
Tel. +34.93.414.53.47
Fax.+34.93.209.11.09

AVISO DE CONFIDENCIALIDAD: Este mensaje y sus documentos adjuntos, pueden contener información privilegiada y/o confidencial que está dirigida exclusivamente a su destinatario. Si usted recibe este mensaje y no es el destinatario indicado, o el empleado encargado de su entrega a dicha persona, por favor, notifíquelo inmediatamente y remita el mensaje original a la dirección de correo electrónico indicada. Cualquier copia, uso o distribución no autorizados de esta comunicación queda estrictamente prohibida.CONFIDENTIALITY NOTICE: This e-mail and the accompanying document(s) may contain confidential information which is privileged and intended only for the individual or entity to whom they are addressed. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution or use of this e-mail and/or accompanying document(s) is strictly prohibited. If you have received this e-mail in error, please immediately notify the sender at the above e-mail address.