Cozmo Robot for Computer Science Education

Faculty Development Competitive Research Grant Program 2018-2020

Purpose:

Using robots to teach computer science (CS) is common world-wide at the primary and secondary school levels, and at the early undergraduate level as well. Devices that manipulate the real-world environment through a simple programming interface help students understand abstract programming concepts by observing concrete actions. Robots in CS are one part of a larger effort to make computer science more attractive and accessible to younger students.
One of the most common robot platforms for education is the Lego Mindstorms [1] robotics kit. It is an excellent platform for young students, and a reasonable platform for freshman college students, but the Mindstorms platform is not suitable for more advanced computer science students. It cannot be used to explore interesting problems such as scene understanding of a robot’s environment or the self-organizing behavior of multiple mobile machines.
With the introduction of the Cozmo robot (by Anki, Inc., USA) [2] in 2016, it has become possible to use the same platform at multiple levels of education. The Cozmo robot is, in many ways, a more simple robot than the Mindstorms robotics kit, but it importantly includes a camera and image recognition software. It also costs half the price of a Mindstorms kit. The figure below shows the Cozmo robot with its “toy blocks” that it can recognize and move around.

In this project, we propose to use the Cozmo robot to enhance computer science education at multiple levels, with the following three objectives:
1. Develop and evaluate Cozmo-based computer science education materials for primary and secondary school students.
2. Develop and evaluate new Cozmo modules for our first year computer science courses at NU.
3. Use the Cozmo robot for undergraduate research projects for third and fourth year students.

Novelty and Significance:

This grant aims to integrate the same robot into the computer science curriculum at multiple levels, ranging from using it to teach basic programming concepts to children, to using it as a platform for research in computer vision and machine learning for upper-level undergraduate students.

Impact:

This project will introduce and evaluate new ways of teaching technology concepts to children, and, consequently, in accordance with the new government program “Digital Kazakhstan”, help to create a larger proportion of Kazakhstani citizens who understand how to create technology, rather than just use it. The curriculum materials from this part of the project will be disseminated to primary and secondary school teachers throughout Kazakhstan. The use of a new, more versatile, and significantly less expensive robotics platform than those currently used in secondary school education will put the opportunity to work with such technologies into many more children’s hands.

At the undergraduate level, computer science students at NU will get experiences working with programmable robots that were previously not available to them, and opportunities to do applied research in the areas of computer vision and deep learning, which are arguably the most popular topics for our students .

Previous Works:

Using robots as a means to attract students to computer science and to teach computer science has been investigated extensively. There are numerous educational robots designed to teach young people about computer science and robotics, such as Sphero [3], Mindstorms [1], WeDo [1] and Finch [4]. Despite a large study by Fagin, et. al, [5] in 2003, that showed slightly negative effects from using robots in a first year computing class at the US Air Force Academy, efforts in the area have increased steadily. Fagin and his colleagues surmised that some of their negative results were caused by the students not having enough access to the robots to complete their exercises, not having access to a robot simulator, and the inexperience of some of the instructors. As further research has shown, such problems are easily overcome. Lauwers (now of BirdBrain Technologies) and Nourbakhsh of Carnegie Mellon University (CMU) have done significant work in the area [6-8], and recently, Miller, Nourbakhsh and Seigwart published a comprehensive survey of robots for education [9]. Current research shows that young people enjoy working with robots, and the robots help them make some abstract ideas from computer science more concrete.

The U.S. National Science Foundation has funded many projects on robots for education; a significant example was the ARTSI (Advancing Robotics Technology for Societal Impact) Alliance [10]. A large, multi-year, collaborative grant, ARTSI’s main focus was using robotics to increase the computer science pipeline. Dave Touretzky of CMU, who was a PI of the ARTSI grant, has also done recent work on computer science education with robots [11,12]. Touretzky was an early adopter of Cozmo, and has done a significant amount of work with it in the last year, for education of young children and for teaching robotics to upper-level undergraduate and graduate students (details in the research methods section).