Flexible techniques for fast developing and remotely controlling DIY robots, with AI flavor

During the last years we are witnessing a very successful osmosis between innovative and cost-effective credit card - sized computers and education. These computers, equipped with low cost sensors or actuators, can be the “heart” of various DIY robotic artefacts. This environment allows for a mixture of thinking and making activities that can be very meaningful in terms of pedagogy and science. Indeed, similar practices, usually referred as STEM or STEAM activities, are applied in many educational institutions, from primary schools up to universities, with most of the effort to focus on secondary school students. The overall process, although promising at the beginning, is not always straightforward to keep up with. More specifically, as students get more experience, they develop a hunger for more complicated scenarios that usually demand features like remote interaction with simple Artificial Intelligence – A.I. capabilities or sophisticated control of their robotic artefacts. At this moment, trainers should be able to propose simple and stable techniques to their students for implementing such features in their constructions. This paper proposes flexible methods for this to be done by exploiting the very popular MIT App Inventor and Snap! visual programming environments, in conjunction with a modified tiny web server module, written in Python, that runs on a Raspberry Pi credit card - sized computer. Furthermore, this paper reports on simple techniques being used to make robust enough robots by low cost everyday/recyclable materials like cardboard, wood, plastic bottles or broken toys