A team of scientists at UC Santa Cruz, Stanford and University of Washington is seeking to develop ways for humans and robots to collaborate, surpassing current technology in which a human must give direct orders to robotic technology.
Robotic technology is used in a large variety of applications, ranging from production in factories to dismantling space junk. While robots are already used for various kinds of surgical procedures, current robotic technology is limited to performing tasks which have specifically been assigned to them by human surgeons.
Professor and director of the UCSC bionics lab Jacob Rosen collaborated with a team of scientists including Stanford associate professor Allison Okamura and professor Blake Hannaford at University of Washington on a robotics project. The project aims to develop the potential for robots and humans to work together in tasks such as surgery and manufacturing. A $3.5 million grant from the National Robotics Initiative, a subsect of the National Science Foundation, funded the project.
“It was a very competitive program,” Okamura said. “The national robotics initiative funded only a few large projects and ours was one that was selected.”
Surgery is already a collaborative effort in most cases, but working with a fellow human is different from cooperating with a robot. When two human surgeons are working together, Okamura said. “They’re both standing around the patient. They can communicate verbally as well as through gestures, and they can see each other’s facial expressions. They can feel each other’s forces while operating within the body.”
Current surgical robots are much easier to control than the proposed intelligent robots that would be able to make decisions and collaborate with a human partner.
“Robots used today are easy to control because they aren’t intelligent,” Hannaford said. “They’re just following a surgeon’s motion.”
Surgeons collaborating with robots have previously been proven difficult because collaboration must involve more detailed and precise communication than modern robotic technology necessitates.
“As robotic functions get more intelligent,” Hannaford said, “the surgeon has to be able to communicate a task very clearly to the robot that involves a lot of spatial things such as locations, trajectories and speeds.”
Presently, robots have a poor physical sense, making it difficult for them to convey information to a human surgeon through physical means, Okamura said.
“We don’t have ways for the robot to communicate well to the human what the robot is trying to accomplish,” Okamura said. “The challenge is coming up with a way of encoding this and building a mathematical and software framework around how robots and humans can communicate physically.”
Hannaford said robots also lack the ability to communicate verbally.
“Human surgeons collaborate by talking,” Hannaford said, “and robots aren’t good at understanding speech.”
While the focus of the project was on robotic surgery systems, the results of the project have widespread application to many fields.
“Given the complexity of the surgical tasks,” Rosen said, “the project may lead to generalization of algorithms developed for collaboration and automation for industrial applications.”
The hardware used in the project was the new Raven II robotic surgery system. The use of the system was crucial to research because it is open-source and enables scientists to build upon the results of others.
The collaborative nature of the research facilitated by Raven II enables scientists to make greater progress than they could by working independently, Hannaford said.
“The labs that have Raven II robots are actively developing new software and testing it on the robots,” Hannaford said. “Also, people are inventing new hardware that can be added to the robot.”
Okamura said it was particularly important that the project was open-source.
“There’s an incredible advantage to there being a standard platform because it enables researchers to collaborate more easily,” Okamura said. “In this particular project, it’s important that all the sites be able to access each other’s research.”
The team plans to make the Raven II technology available to more research campuses. Rosen and Hannaford are starting a company called Applied Dexterity, Inc. that will sell Raven II systems and research.
Scientists face a myriad of issues in developing robots which can act as partners rather than subordinates. Okamura compared the difficulties of human-robot collaboration to the skills necessary for partner dancing.
While dancing, Okamura said, “You apply pressure to each other, but you both already know the dance and hear the music — it’s based on a priori knowledge.”
