Imagining the Robots of Tomorrow 

Twentieth-century film and television are chock full of depictions of robots that tell a deeper story about the relationship humans have with technology. From the fears stoked by Cold War era advances in tech, to space travel that was amplified through the aid of droids and cyborgs, society has had a lot to say about the power of robots.  

And we’ve been especially obsessive about cautionary tales. In the 1980s, Arnold Schwarzenegger’s famous killer cyborg in The Terminator told of a future in peril due to advances in artificial intelligence. The 2008 Disney Pixar film WALL-E told the story of a solitary robot left to clean up an uninhabitable planet earth after humans failed to preserve it. 

As bleak as some of the stories are, it’s not all bad.  

Some takes in science fiction have shown just how far advances in areas like biomedical engineering could go because of robotics. For instance, the world of Wakanda, the fictional afro-futurist country set in sub-Saharan Africa in the Marvel universe, has scientists who have managed to create technology that is controlled through neurological commands. The biomedical engineers of Wakanda have even been able to build virtually indestructible prosthetics using rare minerals.  

Not all of us, however, have access to space-age metals that don’t degrade. At Duke, researchers are trying to figure out just how practical robots can become in our everyday lives. 

And it starts by asking the right questions.               

Question one: How do we build general purpose robots? For Boyuan Chen, the answer goes beyond focusing on a singular application. As an assistant professor in the Thomas Lord Department of Mechanical Engineering and Materials Science (MEMS), Electrical and Computer Engineering (ECE), and Computer Science, Chen’s research is more than just the study of robotics. His specific interests are in developing “society-centered generalist robots” that learn and improve through dynamic interactions with the world, aiming to address significant questions in our society. 

“One unique aspect of my research is to seek concrete but fundamental principles that can power the development of intelligent machines for diverse domains and applications,” Chen explained, as he guided me through a winding corridor of robotics labs at the university. “Through our work, we study the science of embodied intelligence to answer what are the basic ingredients that we can bake into machines such that they have the capacity to learn and evolve over time. 

Chen is the director of Duke General Robotics Lab, which is broadly interested in both the hardware and software development of AI-enabled robotics, which may cover fields like robot learning, perception, machine learning and human-AI teaming. 

The hope is that, through this dynamic engagement, these robots would ultimately acquire high-level cognitive skills by themselves to work alongside humans as our partners or teammates. Chen upends the conventional approach to thinking about artificial intelligence by instead realizing that machine intelligence has to have a “body and brain.”

From mechanical systems to software and algorithms, Chen says the connective tissue—the “Holy Grail” of all of this—is adaptation. “You have to build mechanical systems with sensors around them, but you also have to look at software, which we call the brain, and its algorithms, for example, to build out a substantial perception system.”  

Though a relatively general term, adaptation in Chen’s eyes should not be constrained to a narrow domain. Oftentimes people may think adaptation only refers to the way humans do it—we see data sets, perform tasks in various environments, and adapt where and when necessary. When we enter a new environment, we can quickly pick up new skills or adapt to that new environment, like learning to balance on icy surfaces. 

Chen takes this a step further, understanding that we can pass on this adaptation skill to robotic systems. “They may not have the same sensing system humans have, but other robots with different body forms and configurations may be able to acquire the same skills,” Chen explained.

Preparing Future Robotics Specialists

Question two: Can robotics improve the already impressive abilities of our finest surgeons? Duke’s National Science Foundation Traineeship in the Advancement of Surgical Technologies (TAST) program is training graduate students with that very goal in mind. 

TAST combines multiple disciplines to design advanced surgical technologies that consider provider, societal, end-user and patient needs in their development and testing. The program provides a pathway for graduate students in engineering and computer science to design fundamentally new technologies to ultimately advance medical practice. 

“The reason why medical robotics is so important is because, when you want to do anything with medicine, it gets increasingly more complex and nuanced,” said Siobhan Oca, assistant professor of the practice in MEMS, who has been instrumental in designing the new curriculum alongside fellow MEMS professor Brian Mann. One of the strengths of TAST is the opportunity to collaborate with the breadth of medical faculty at Duke.  

Through this traineeship, fellows engage in research goals that build collaborative relationships while developing professional skills through focused group activities. And to augment the program, multiple certificates in robotics have been implemented for students to engage with these convergent fields.  

The Surgical Education and Activities Lab (SEAL) is another resource with medical robotics in mind. This lab is a state-of-the-art surgical simulation center designed to provide advanced and innovative training in a risk-free environment. Oca says preparing students for industry is a significant feature that shouldn’t be overlooked. 

“Conversations about what the industry offers and is looking for really empowers students to think about where they can be part of the robotics workforce in a thoughtful way,” Oca shared. “You are looking at jobs in particular that are focused on advanced degrees and skills that come from focused research training.”  

Exposing students to these diverse experiences enables them to pursue their specialized interests and offers more career opportunities to create the robo-surgeons currently relegated to science fiction. 

When Robots, Space and Medicine Collide

Question three: Can robots help us reach for the stars? Scientists and physicians at Duke led by Dan Buckland, assistant professor of emergency medicine, are developing technology for NASA that looks at improving the health and performance of astronauts on exploration missions. 

“From a robotics standpoint, what I look at is automation and, by extension, what it’s going to look like on these exploration missions,” Buckland shared. “I also try to understand how automation in exploration impacts human health and performance in particular.” Buckland also leads the Duke Acute Care Technology Lab (DACTL), where they research developing technology for the diagnosis and treatment of acute diseases using data science and robotics methods. 

Buckland connects these various areas of focus by investigating automation in safety-critical systems, which are systems that, if they fail, will result in loss of life or limb. “There are plenty of things humans do in an emergency room, ambulance, healthcare system or even a spaceship that can be automated,” Buckland explained.  

Discussions about AI are frequently confined to people interacting with devices or computers, but he notes that it becomes robotics when automation in a computational world impacts people without someone facilitating that connection. “I only have so many people or so much space,” Buckland said. “Robotics allows me to deliver care remotely and at a bigger scale.”  

“Duke has several biomedical engineering courses that focus on clinical problems through a BME lens,” Buckland explained. “TAST and its focus on mechanical engineering methods provides a unique perspective that starts with an engineering and automation approach to health problems, instead of clinical problems searching for engineering solutions.”

This kind of medical engineering is different in that it distributes competencies across disciplines. Going into a hospital, Buckland says he is one of a select few doctors who also have expertise in engineering.  

“I’d say one of my goals is to help students become valuable contributors in both fields. You may not go to a hospital looking for the best engineer or an engineering school for the best doctor, but it’s the kind of guidance we offer students looking to master this dual profession,” Buckland said.   

Cross-professional collaboration typifies how robotics at Duke is transforming the field. The TAST program, the university’s robotics lab and the extensive course offerings equip incoming graduate students with the tools needed to navigate multiple disciplines, all while existing at the intersection of medicine and robotics.  

“One of the benefits of Duke is its geography,” Buckland said. “The medical school and the engineering school stare at each other from across the street. The faculty in both go to the same coffee shops and cafeterias, even the same sandwich shops. But what we’re doing is working continuously to bridge that gap so that we’re closer than just across the street at the end of the day.”