Attending ROS Kong 2014

As part of the OPW program with OSRF, I was able to fly out to the University of Hong Kong to attend  the first official ROS users meeting in Asia, ROS Kong 2014 [1]. ROS (Robot Operating System) is an open source framework (a collection of software tools, libraries and drivers) mostly used in robotics. There I met with one of my mentors, Tully Foote,  co-OPW intern Louise Poubel [2] and a subset of the international ROS community in attendance.

Speakers at the event highlighted research in a variety of topics that included service robotics applications, mapping, new and updated ROS packages and experience competing in the 2013 DARPA Grand Challenge. The presentations did not focus on the technical details but emphasized the range of research made possible with ROS. For example, getting a PR2 to fetch a sandwich :) [3]. As an added treat, there was also a demonstration of the HKU Team’s Atlas robot. Atlas first used its cameras and Lidar sensors to grasp a pipe/baton to demonstrate semi-autonomous operation. After a quick calibration dance, the robot proceeded to pick and place an empty 5-gallon water bottle, a hand drill, then quite delicately, an egg.

ROS for Low Cost Robotics in Education

As part of the Birds-of-a-Feather session, there was a discussion on the use of ROS for Low Cost Robotics specifically for use in a budget conscious laboratory. The small ad-hoc meeting led to several threads which included defining what “low cost” means, the robotic platforms available, the materials in the ROS website and the challenge of getting (younger) students started with open source robotics.

Low cost robotics is by no means anything new, since several designs have been developed over the years. The 10 USD Ultra Affordable Robot Design Challenge organized by  the African Robotics Network was notably mentioned [4]. However, to maximize the communication tools and libraries provided by ROS, the robot must run (at the minimum) on a microcontroller board like the Raspberry Pi or compatible Arduino board. This of course, will affect other design elements, such the power supply requirements, network support, sensors and motors. Looking at out-of-the-box robots, there is the go to iRobot Create (22o USD, base only [5]); only make sure to bring your own laptop. Or to get the full experience, there is the Turtlebot 2  (1,995 USD, complete kit [6]) or Aldebaran’s newest robot, Pepper [7] which is said to retail at around 2,000 USD. A laboratory should expect to spend an additional 50% or more on top of the tagged price to  cover shipping and taxes – which places the budget somewhere at the 3000 USD/robot range. Surprisingly, there was no mention of the Lego EV3 or its older brother NXT 2.0.

The hardware here is just half of the equation, since majority of users in Asia are used to Windows PCs and schools are usually on an MSDNAA account, it is difficult to have students embrace the intricacies of ROS and the Linux ecosystem. For example, I remember how a student was sharing war stories on setting up his laptop and desktop for both Windows and Ubuntu so he could do thesis related work with ROS Fuerte on Ubuntu and use Windows for a majority of school projects. A Virtual Machine could have made the setup simpler, but the Kinect refused to play nice. His groupmates were even more reluctant, as they would write their code in Windows and then translate them to Python on Ubuntu to integrate their modules for the project they were working on. While there is an experimental version of ROS for Windows, I was told that the handling of package dependencies was an issue. Apart from the Operating System, it was also mentioned that support for additional languages can also improve learning ROS and the software libraries that come with it.

At OSRF, new efforts in teaching robotics with ROS (with the Turtlebot) are in progress through the work of OPW intern Mariana Liebman-Pelaez. Read more about her work on ROS Courses here: [8].

 

Final Thoughts

ROS Kong 2014 was a look into ROS and its impact on robotics research. The event brought together ROS users to talk about on going projects and share working experiences. A good deal of lessons can be learned through implementing robotic projects. Apart from the mechatronics, theories and algorithms involved, students/researchers alike experience the importance of collaboration, repository development, proper code structure and documentation. It is the balance between tinkering with hardware and software that make robotics enjoyable and appropriate to introduce to kids (and the kid at heart). The work that OSRF and the ROS community has done to streamline robotics software is incredible, evidenced by the number of laboratories and robotics companies who now use and contribute to ROS . In line with the topic that drew me to apply to the OPW, my interest is still with working on tools/materials to get students started and excited to use open source hardware and software in the CS curricula and/or research laboratory.

It was such a privilege to be able to attend ROS Kong 2014. It was a fitting finale to my OPW internship experience. A huge thank you goes to the Gnome Foundation and OSRF for extending this opportunity.