Building Robotic Manipulators

Estimated reading time: 3 minutes

Robots with arms and hands cannot yet handle the unpredictable nature of the world around us, but Matei Ciocarlie is aiming to change that. Ciocarlie’s research focuses on building smart devices that can interact with their environment—whether it’s helping a stroke sufferer regain hand mobility through robotics or creating a robot to perform complex manipulation tasks.

“Chances are most people have never seen a robotic manipulator because they’re mostly in factories,” says Ciocarlie, assistant professor of mechanical engineering, who joined Columbia Engineering in 2014. “They do the same task again and again, very precisely and tirelessly. Some of our research aims to give robots similar capabilities, but in very different domains.”

“In inhospitable environments,” he continues, “we are getting closer. In a few years time, we could have robots for disaster response. The home, though, is the toughest of environments in some senses. Having manipulators that can actually deal with typical clutter in someone’s home is still far away.”

Ciocarlie envisions robotic manipulators—those with hands and arms—helping in a variety of situations, such as aiding in catastrophe scenarios, like earthquakes or hurricanes, and assisting people with motor impairments, such as stroke or accident victims.

He is working closely with Joel Stein at Columbia University Medical Center on a project to determine how robotic manipulation technology can benefit stroke victims who have lost their grasping and control abilities. The two are hopeful their technology will help people with other degenerative conditions that affect manipulation.

“Robotics is squarely at the intersection of mechanical engineering, electrical engineering, and computer science,” notes Ciocarlie who directs Columbia’s Robotic Manipulation and Mobility Lab, or ROAM—an acronym he coined to denote that robots are free and interact with their environment. “One of the key challenges in robotic manipulation is how do you reduce the complexity of the problem without losing versatility?”

To this end, he is developing a robotic hand that’s as versatile as the human hand, able to perform diverse tasks. Ciocarlie admits the technology isn’t there yet, but researchers in his lab are making headway.

Recently, the Office of Naval Research named Ciocarlie a 2015 Young Investigator Program recipient, awarding him a three-year, $600,000-plus grant to research human-in-the-loop systems. His team will study humans and robotic manipulators working side-by-side on the same task, such as routine maintenance and disaster response.

“We aim to build the artificial hands and control interfaces that allow the person to manipulate the environment through the robot, while leaving the human's own hands free to engage in the task as well,” Ciocarlie explains.

Ciocarlie’s interest in robotics started after he earned his bachelor’s in computer science in Bucharest. He quickly realized that, when applied to robotics, the results of computer programming lead to tangible, physical manifestations and affect change in the real world. Ciocarlie continued his education at Columbia Engineering earning both his master’s and doctorate at SEAS while being guided by Computer Science Professor Peter Allen. His PhD dissertation focused on the computational complexity associated with dexterous robotic grasping—a multifaceted topic he continues to research today.

“We’re going to build robots that are more intelligent by developing algorithms, or the brain, together with mechanisms, or the body,” he says. “Columbia is well positioned to take this approach.”

Prior to joining Columbia, Ciocarlie was a research scientist and group manager at Willow Garage, a personal robotics applications developer in Silicon Valley, and later worked as a senior research scientist at Google. In 2013, Ciocarlie was awarded the IEEE Robotics and Automation Society Early Career Award.

He observes that while robots might be required to do different tasks in varied scenarios, many of the underlying skills will be the same, and similar principles apply.

“Any technology that gets developed needs to make somebody’s life easier and needs to have an impact,” says Ciocarlie. “Ultimately, that’s the robot I’m looking to build.”