Mechanical Engineering Assistant Professor Xiaoli Zhang has received an NSF CAREER Award for her project to improve the human-robot control interview during remote operation.
With a novel goal-guided control interface, instead of passively following an operator’s motion input, the robot will understand the operator’s high-level objective during an object-grasping operation and autonomously conform to task constraints to reduce control difficulties and ensure the success of subsequent manipulation.
Challenges in teleoperation include indirect visualization and manipulation, as well as the discrepancy between the robotic grip and the device being used to manipulate it.
“Even using a data glove, the physical structure of a human hand and the robot’s hand are extremely different. If we want robots to do fine manipulations like a human can, we must solve this control problem,” Zhang said.
One of the most successful implementations of robotic teleoperations is in surgery, where it might be used to remove a gallbladder. Teleoperations are also extremely helpful in repairing or inspecting mines, space exploration, search and rescue operations and anywhere that is difficult or dangerous for humans to access.
“Ultimately, if we want a robot to think like a human, to be intelligent and have autonomy and the capability to regulate itself in order to work with humans, the robots have to first become more aware of how humans achieve those things,” Zhang said.
“Even when it comes to something as simple as picking up a cup, there are multiple ways we approach it, depending on whether our goal is to pass it someone else, place it on a shelf, drink from it or wash it,” Zhang said. “We have to investigate our own behavior patterns in order to formulate a knowledge-based model through machine learning methods for a robot.”
Zhang also seeks to improve distance learning by developing a system that immerses remote students in the classroom by allowing them to control a robot for object manipulation or interaction with classmates.