Q&A with Matei Ciocarlie
October 13, 2017
Columbia Technology Ventures (CTV) spoke with Matei Ciocarlie, Assistant Professor of Mechanical Engineering, to learn about his efforts to create robotic hands capable of healing—and helping.
CTV: There is tremendous commercial potential for dexterous robotic hands—why is it so difficult to design and create them?
MC: The best way to appreciate how difficult robotic manipulation can be is to consider your own hand. With more than 20 joints, more than 40 muscles, and incredible tactile sensors in each fingertip, the human hand is underappreciated as a piece of hardware. Between this physical hardware and the software that controls it—the brain— a human hand is an incredibly complex, remarkable piece of machinery.
When we look at the state of robotic manipulation today, there are very few instances of robots interacting with objects in ways that resemble how humans interact with them. The most common use of robotic arms and manipulators is for assembly, and those robots are fed the same part in the same way again and again. They’re tireless and precise, but that’s the only way they can operate. Robots in the home and in service domains are becoming more prevalent, like delivery robots and those in hospitals and hotels, but their duties are mostly limited to moving objects that have been loaded onto the robot by a human—they don’t typically have hands and they don’t engage in versatile manipulation. There’s such a big need for this, but replicating the feats of the human hand is incredibly difficult.
CTV: How are you attempting to tackle the problem?
MC: We are working on both the hardware and the software for manipulation. We are building new types of tactile sensors that aim to recover orders of magnitude more data than existing technologies. We are designing robot hands combining active control with passive compliance. And we are studying the fundamentals as well, looking to improve the theory of manipulation. These are all bringing us closer to the goal of making robotic manipulation more versatile, more capable of interacting with complex objects in a wide range of shapes, settings and scenarios.
CTV: Tell us about how you’re applying these findings to unmet needs in rehabilitation medicine.
MC: Nearly one million people in the United States alone have strokes every year—75 percent of patients have their arm or hand affected, and half don’t fully recover limb function. We have an incredible collaboration with Dr. Joel Stein, chair of the Department of Rehabilitation and Regenerative Medicine at Columbia University Medical Center, and through our work together, we’ve seen the huge need to provide assistive technology for these patients to engage in versatile manipulation.
Studies have shown that more training is highly beneficial for stroke patients, but access to occupational therapy services is often limited. Dr. Stein and I set out to create a wearable device that could assist hand movement and enable patients to improve their strength and dexterity at home, rather than being dependent on a hospital or clinic for therapy. We co-created the MyHand orthosis, which fits over the hand like a glove and allows for versatile manipulation. We’ve been testing prototypes with stroke patients at Columbia for more than a year, and the response has been both remarkable and very helpful in improving the concept.
CTV: Does this bring you any closer to solving the problem of creating hands for industrial robots?
MC: We’ve learned some surprising things about translating the complexity of human grasping to machines. For example, we've seen that, while the human hand is very complex, we can build prototype robotic hands that are quite dexterous with few actuators. We’re a long way off from robotic manipulators that can operate in a typical home (empty the dishwasher or make a sandwich), but this kind of research could help can have immediate impact in fields like logistics, and e-commerce.
We’re working on creating robotic hands capable of a task we refer to as “manipulation in clutter”, or the ability to reach into a bin of loose parts and identify a single, specific item. It’s something that the human brain and hand can easily manage, but robotic hardware and software haven’t yet achieved. We’re developing technologies to enable robots to handle individual items, and for fields like materials handling, logistics, and e-commerce, anything that improves the flow of items stands to make a big impact. Think about bin-picking tasks—every time an order is placed on an e-commerce site, the selected items have to be located and picked from a warehouse and packed for shipping. These tasks are highly repetitive, error-prone, and bad ergonomics can easily lead to injuries, so there’s certainly a place for robotic assistance.
I also want to mention that this work would be impossible without the breadth of knowledge and range of resources here at Columbia. Our research requires input from many disciplines—electronics, materials science, robotics, data science—and we have put together an incredible team to collaborate on these projects.
Columbia Technology Ventures works closely with Columbia researches to commercialize early-stage technology innovations, connecting industry and investment partners with researchers to bring impactful, in-demand robotic technologies to market as quickly as possible. To see Columbia’s Robotics portfolio available for licensing click here: http://innovation.columbia.edu/robotics
Please send all inquiries to: firstname.lastname@example.org