How Robots Learn: A Video Series

IEEE Transmitter features bots that play soccer and work in hospitality

13 November 2017

For the past few months, IEEE Transmitter has been posting a series of videos and articles that show how robots learn to accomplish specific tasks. They include a 3D-printed robotic arm that can communicate in sign language and a robot that learns to grasp objects, like cups and books.

IEEE conducted interviews with robotics experts who have shared demos of the robots they work with. Watch the videos below to see a few of them in action.

  • Scoring Goals

    In this video, IEEE Member Esther Luna Colombini discusses soccer-playing robots that she has been building. She is a professor of robotics and artificial intelligence at the University of Campinas, in Brazil.

    Colombini’s autonomous androids are designed with cognitive systems that can identify a ball; make an informed decision on what to do with it, such as pass it to a teammate; and learn to improve their game through practice. The robots are nimble and built to imitate athletes’ movements such as running and getting up after they fall.

  • Helpful Humanoids

    Asimov Robotics’ humanoids can work alongside medical professionals to help patients, or assist customers and hospitality employees with a variety of tasks. The company’s founder, IEEE Life Member Jayakrishnan T, describes in this video how its robots learn to process social cues so people are more comfortable talking to them. The company customizes the robots’ responses depending on the environment they will work in.

  • Getting Around

    Whether cleaning a house or driving people to their destinations, autonomous systems must learn how to get around without running into obstacles. In this video, IEEE Member Ming Liu talks about his work in teaching robots to navigate their surroundings. He is an associate professor of computer and electrical engineering at the Hong Kong University of Science and Technology.

    Liu is researching two types of learning styles: real-time learning, in which the robots figure out where to go in the moment; and a learning-based method, whereby the robots get more practice moving around their environment and begin to make smarter decisions, such as figuring out shortcuts to a destination.

Learn More