Coding the iCub – Making Connections with Cognitive Robotics

An international partnership between University of Lethbridge neuroscience professor Dr. Matthew Tata and Dr. Francesco Rea (Italian Institute of Technology in Genoa) has given three undergraduate students the unique opportunity to work on an independent study focused on cognitive robotics.

Tata started working with Rea during a sabbatical year in 2013-2014 when he lived in Genoa, Italy and worked full time at the Italian Institute of Technology (IIT).

“It was a perfect example of how a sabbatical study leave should work. I spent a year working with Dr. Rea learning everything I could about robotics, which was completely absent from the curriculum and research environment at the U of L. I brought robotics back with me and set up a lab and an introductory course, and maintained the collaboration with Dr. Rea so that students and researchers at the U of L can tap the expertise of IIT,” says Tata. “Our collaboration with colleagues at IIT aims to translate our discoveries about human cognitive neuroscience into technologies for robot perception and attention. This will make robots better at understanding and responding to human instructions.”

Rea received a BSc in Software Engineering from the University of Bergamo, Italy and went on to do an MSc in Robotics and Automation at Salford University, England. He then received a PhD from the University of Genoa in 2013 where he developed machine vision and attention algorithms for human-robot interaction. Since 2013, he has been a researcher at IIT where he takes a brain-inspired approach to develop advanced cognitive robotics for the iCub humanoid robot.

“My research is focused on human-robot interaction and also cognitive robotics. These two disciplines work together, and we strongly believe that if we want to interact with robots, we need to share the same neuroscientific models of interaction,” says Rea. “We know that the U of L is a very important university where research in the field is well-known across the world. We are convinced that by studying those models here, we can more easily port them on a complex humanoid robot such as the iCub.”

The iCub is the humanoid robot developed at IIT as part of the EU project RobotCub and subsequently adopted by more than 20 laboratories worldwide. It has 53 motors that move the head, arms and hands, waist and legs. It can see and hear, and it has the sense of proprioception (body configuration) and movement (using accelerometers and gyroscopes).

“The humanoid robot, iCub, is complex – and there is a very specific reason. We want to create the perfect platform for the study of humans, so we need a little bit of complexity that is justified by the fact that we really want to reproduce some human behaviours,” says Rea. “It is complex, but there is a huge community of iCub users that collaborate together. By doing so, we can easily build on each other’s work and quickly provide very new innovative achievements, which relaxes the complexity of the platform a bit and is definitely one of the advantages of working with the iCub.”

Rea was in Tata’s lab working with computer science undergraduate students Lukas Grasse, Austin Kothig and Alex Hochheiden on the iCub robot. The students have been focused on various applications involving the implementation and improvement of cognitive models of the human attention system. He is working on introducing the students to developing new algorithms on a complex humanoid robot, which requires very specific procedures to create new behaviours and skills. “The idea is to achieve cognitive robotics and apply neuroscientific plausible models to robots.

The goal is to improve and validate those models – at the same time trying to give the robot skills that are human-like, skills that will eventually allow the robot to interact with humans in a natural fashion,” says Rea.

The students found their work with Drs. Tata and Rea to be very rewarding, and gained a lot from working with the iCub. Alex Hochheiden has been getting up to speed working with the robot for the first time.

“Other team members have been working with it for a semester or more, so I'm just trying to learn what they know and how I can contribute,” he says. “In addition to that, I've been looking at how we can optimize our code, helping troubleshoot issues and thinking about how to apply some software engineering best practices.”

Lukas Grasse has been working on the iCub’s understanding of speech in the environment and its response to when someone addresses it with “Hello iCub” or in Italian “Ciao iCub”. Like the others, he was first introduced to robotics through an Introduction to Robotics course offered by Tata, and has pursued the interest ever since.

“I really liked the class and have been doing independent studies in Dr. Tata’s lab since then. Cognitive robotics is an awesome field because it requires knowledge from many different areas, and you get to work with people that have different backgrounds and experience,” he says. “The field of cognitive robotics is just getting started. Working with Dr. Tata and Dr. Rea has made me realize how important it is for people to be able to interact with robots naturally using speech, touch, etc. I think this is true for technology in general, and I plan to keep working on solutions that enable these natural interactions.”

Austin Kothig also has plans to continue working with robotics.

“I've been spending a lot of time working in the lab with Dr. Tata and Dr. Rea, and have definitely learned a lot about how the global iCub community develops software for the robot,” he says. “Prior to doing my summer independent study I thought that I would be happy with a Bachelor in Computer Science. Since then I have been inspired to continue on to apply for a master's program doing research on AI and robotics.”

It is important to consider the impact of cognitive robotics in a world where humans have been building machines to help us do work for thousands of years.

“Now, with the advances of computer science, engineering and cognitive science we can build machines that are interactive and smart. They can make decisions. This is the most important technology of the 21st century and it will be highly disruptive,” says Tata. “There are many important technical and societal implications that we need to carefully consider and understand, so Canada needs to have highly-trained scientists working on the state-of-the-art in the field. That means working at the interface between neuroscience and robotics.”

Connect with Dr. Matthew Tata to learn more, or visit his lab website: tatalab.ca