MIT researchers make their automata more delicate in contact with humans by anticipating their possible movements.
MIT researchers are working to improve the physical interactions between their automatons and humans. They instilled in their algorithms the variety of human behavior to make effective and safe contact possible. They demonstrated with a robot helping a person put on the sleeve of a jacket.
The difficulty is to find the right balance between a robot that is too abrupt, which could cause injuries, and one that would not dare to move for the same reason, explain the authors of the research. Instead of limiting themselves to a single default model, where the robot incorporates only one potential reaction, they gave the machine an understanding of many possible reactions to more faithfully mimic the human. As the robot collects new data, it is supposed to reduce uncertainties and refine the efficiency and delicacy of its task.
In addition, the researchers also plan to focus on the study of the subjective feeling of security of the human being in contact. These machines could eventually find an outlet by helping people with disabilities or reduced mobility.