This project strives to develop an inertia compensator for the InMotion2 planar robot for use in studies with human interaction.
The InMotion2 robot from MIT’s Newman Lab is a 2-degree of freedom robotic arm used for studies of human interaction. A human subject grips a handle at the end-effector and is able to push the handle to in the horizontal plane. Two highly back-drivable brushless motors drive the “shoulder” and “elbow” joints of the robot. The motors can be actuated using simple impedance control schemes to force the end-effector within a certain trajectory to imitate constrained motion. In this mode, the human subject can move the handle freely along the path, but cannot deviate from that path. This control scheme is used by the lab in studies of human motor control and can be easily adapted to any sort of related experiment.
A major drawback of this setup that has been observed in human experiments is that along the constrained path (or any unconstrained path), friction and inertia vary based on the configuration of the robot. This makes it difficult in lab experiments to determine whether any changes in velocity of the trajectory are due to the human’s motor control or to the changes in physical friction and inertia. The goal of this project is to develop a controller for the robot that manages the apparent inertia at the end-effector so that the human user feels a constant inertia as they move the handle around the workspace. This is possible using the two torque-controlled motors, encoders on both motors, and a 6-DOF force transducer at the handle. A modified nonlinear adaptive sliding mode controller is used to accomplish this task.