Simon was developed at Georgia Tech’s Socially Intelligent Machines Laboratory by Dr. Andrea Thomaz to study Human-Robot Interaction, such as imitation learning and human-robot cooperation. Since robots cannot be programmed to do absolutely everything before entering the real world, they must be able to learn new tasks as they encounter them. Dr. Thomaz was recently selected as one of MIT Technology Review’s 35 top innovators under 35 for her work in machine learning – algorithms which she bases on human learning mechanisms – that help to give robots like Simon this crucial ability.
Simon uses computer vision and speech processing to understand what its human instructor is doing. While Simon’s head was developed with the help of a designer, its articulated torso, compliant 7 DOF arms, and dexterous hands were developed by Meka Robotics (started by former MIT grads). Simon’s head is child-like in appearance, and also sports two LED-laden ears which can rotate and change color, in order to give the impression of a child-like or non-human intellect while avoiding the pitfalls of the uncanny valley. A video and more after the break.
All told the upper torso is about the same size as a 5’7″ woman, so that human instructors won’t feel threatened or intimidated standing next to it (unlike the rather large Iowa State University humanoid robot). The robot can communicate with a human through its eye gaze, gestures, and by nodding or shaking its head. Only recently completed, already Simon has been taught how to pick up objects and place them inside a basket, and to assemble a toy house among other things in computer simulation. Soon, Simon will get his eyebrows and lips (already they are experimenting with magnets under the face shell to actuate these parts).
- Georgia Tech’s Socially Intelligent Machines Lab (EN)
- Meka Robotics (official site EN)
- Dr. Andrea Thomaz TR35 Article
- Dr. Thomaz interviewed by Robots Podcast
Yvonne Boyd (MIT TR)