My Robot Can Play Football and Do Sit-ups!

​​Can you imagine combining what you study with your favourite hobby?



My dream would be to be able to work on something that involves both my study career and my greatest passion in life that is football. Although my study program is a bit distant from what I love most, here in Chalmers, I have had the opportunity to combine both. 

First, I took the course of intelligent agents where we had to develop an application based on human-machine interaction. This application had to include image processing, speech synthesis, voice recognition, and internet search to create an assistant that will help you in a specific task. Similar ideas to Siri, Alexa, and Cortana, but, for this case, it should be built for a specific task and not as extensive as those assistants. I did not hesitate twice, and I proposed to my professor an assistant to help you with the statistics of your favorite football team. 

The main idea of the project was that you could start a conversation with the agent as if you were talking with a person. Therefore, I proposed different data that the agent could handle so that you could have a normal conversation with him (I named it Fabio, which makes it a He :)). Some of the data that Fabio can provide you with are: the next match, position in the league, difference of points with some other team of the same league, classification to Champions league, the result of the previous match, among other functions. The agent can look into three different leagues: Bundesliga, La Liga, and Premier League.  I had to decide between only some of the most important leagues since, at the time the agent does an internet search, it would be more complicated if it tries to search all the leagues in the world.

The second chance I had was in the humanoid robotics course, where I created a project in which a Bioloid robot could perform a task designed by us. I convinced my team to design a penalty shooter (cool right? :)). We had to take into account several limitations due to the time we had to develop the programming of the movements. So, we decided to have no goalkeeper and a fixed position for the ball and the goal. After that, we had to teach the robot to stand up, which is already quite challenging on its own, and finally, do the kick by directing the ball either towards the centre, left or right. What it means that the challenge is not only to stand up but to accommodate the body to give direction to the ball. The method we implemented was to design the movements in a 3D simulator to have a model to follow. Thereafter, using deep learning, a neural network was designed to teach the robot from scratch the movements that should be made to make the kick.

All in all, the experience I had during my studies at Chalmers has been quite rewarding for me because it has allowed me to combine my passion for football and my knowledge, which I hope will be the same in my future work :)



Author: Dante

Published: Mon 17 Feb 2020.