Ming-fai Fong- Designing an intelligent walking “toddler” robot by Melis A. '08
After learning to sit, bend their knees, squat, and stand, babies can finally gather the coordination and strength to walk. While engineers have mastered robotic arm designs, walking robots have yet to even come close to mimicking the human gait. In an attempt to make a robot that imitates humanlike walking as much as possible, the researchers in the Seung Lab designed a robot that learns how to walk in the same way that humans do.
Parents always eagerly anticipate their child’s first steps and cherish the memory for the rest of their lives. Senior Ming-fai Fong, a Mechanical Engineering student with a concentration in Artificial Intelligence, experienced some those feelings even before becoming a parent, because of her work in the Seung Lab at MIT to develop the mechanical design of an intelligent walking “toddler” robot.
After learning to sit, bend their knees, squat, and stand, babies can finally gather the coordination and strength to walk. While engineers have mastered robotic arm designs, walking robots have yet to even come close to mimicking the human gait. In an attempt to make a robot that imitates humanlike walking as much as possible, the researchers in the Seung Lab designed a robot that learns how to walk in the same way that humans do. Ming says, “The idea is that babies are equipped with muscles that they need to walk, all they need is a little bit of control in how to use these muscles.” By creating a mechanical robot with a controller that uses a learning program, the robot can learn to walk after about 600 steps.
Ming came to MIT with a strong interest in robotics and artificial intelligence. She found out about this UROP opening through the UROP website and was surprised when she was offered the position. Her mentor, Russell Tedrake, was an MIT Computer Science graduate and was looking for a Mechanical Engineering student to help him out. Ming said that working in interdisciplinary research, with a computer scientist mentor in a brain and cognitive sciences laboratory, was one of the most exciting aspects of her research. She worked on the project from the beginning of her sophomore year continuously through the beginning of her junior year. Her research involved improving the controller and experimenting with different mechanical designs. The version of the robot that she worked on did not have knees but had a simpler controller and since then two other students have improved the robot.
Ming has also had UROPs in the Media Lab, a Chemical Engineering lab, and now she is working in a cognitive science lab. She came to MIT to do research and clearly she has gained plenty of experience throughout her four years here. Ming says that doing research is the best way to learn and that the wide variety of research opportunities is what makes MIT unique. She would like to eventually go to graduate school to pursue rehabilitation-related engineering, but she doesn’t know if she first wants to work as a research assistant in the US or to travel to Mexico and develop software for distance learning through an MIT iCampus project.
That’s awesome — do most people entering the UROP program have background in what they go into or are they taught what they need to know?
It depends on the UROP. No matter what type of UROP of you do, you’ll have a lot of on the job training. Some UROPs are primarily for upperclassmen who have taken classes (such as biology labs) that give them the background that’s essential for a successful UROP.
hi, its harash i always think about the wonderful world of robots & i think that i cn get some ok with ur company…….
will u like to tell me more abt ur creation….
i liked you choice of designing a toddler concept bot.so in the leaning the time is a factor?.is the bot follows a algorithm to ;leans or uses its mistake or what dfoes it follows.i would love top know about it.
best of luck