MIT Department: Mechanical Engineering
Undergraduate Institution: Yale University
Faculty Mentor: Neville Hogan
Research Supervisor: James Hermus
My name is Michael West and I am from Rancho Cucamonga, California. I am a rising senior at Yale University majoring in mechanical engineering. My primary interest of research is the mechanical design of medical devices and robotics. Following graduation I aim to pursue a PhD with a focus in robotics. Other than being an engineer my hobbies include Netflix, card games, and pick up basketball.
2017 Research Abstract
Analysis of Human Control in an Effort to Improve Robot Dexterity
A. Michael West Jr., Department of Mechanical Engineering, Yale University,
James Hermus, Department of Mechanical Engineering, Massachusetts Institute of Technology,
Meghan Huber, Department of Mechanical Engineering, Massachusetts Institute of Technology,
Dagmar Sternad, Departments of Biology Electrical & Computer Engineering and Physics, Center for Interdisciplinary Research of Complex Systems, Northeastern University,
Neville Hogan, Departments of Brain and Cognitive Sciences, Mechanical Engineering, Massachusetts Institute of Technology,
Human neural pathways and muscles are orders of magnitude slower than computer processors and robotic actuators, yet humans outperform robots in tasks requiring agile or dexterous movements. Simple everyday tasks requiring physical interaction, such as opening a door or turning a crank, present complex dynamic challenges. However, humans can do these tasks quickly, and subconsciously. One of the well-established approaches to manage physical interaction is compliant force control. In this work, an experiment was designed to test the hypothesis. Hypothesis 1: Humans use force control to manage physical interaction during a robot following task. Using the Haptic Master, a handle was moved along an elliptical path, to which subjects were instructed to apply a constant force of 5 N in the direction of the robot’s motion. The handle was moved with a velocity profile in accordance with the two thirds power law. Additionally, in alternating blocks, subjects were given visual feedback informing them of their tangential force. Large position dependent errors in tangential force were observed. These large errors are inconsistent with our hypothesis and suggest that subjects are not using a pure force controller to perform the task. Future work will further investigate motor planning strategies consistent with these observations in constrained motion tasks.