WEAVING HUMAN-COMPUTER INTERACTION INTO THE FABRIC OF ROBOTICS
Autonomous machines that interact, collaborate and co-exist with humans are becoming more and more functional, making their way out of research labs and into industry. But despite successes in robot function, robot interaction with people is lagging behind.
Anca Dragan, Assistant Professor at UC Berkeley, wants to change this and weave interaction into the very fabric of robotics: “I envision riding in a self-driving car as it is effectively coordinating with other drivers on the road and with pedestrians. I envision people with disabilities seamlessly operating assistive devices to thrive independently. And I envision collaborative robots in the home or in the factory helping us with our tasks and even gently guiding us to better ways of achieving them.”
She is also founder of the InterACT Lab, where they are working to enable robots to work with, around, and in support of humans. Focusing on algorithms for human-robot interaction (HCI), Anca and her team are working to improve robotics for different applications, including assisted living, manufacturing and autonomous vehicles.
I asked Anca a few questions to learn more about her work in robotics and the advancements needed in HCI.
What do you find exciting about your current role?
So many things! First, the research that we get to do. It is very exciting when robots that use the algorithms we develop synthesize behavior from scratch that is transparent to people and that improves coordination with them in day to day tasks. When it comes to interaction, so much of the behavior is handcrafted right now, that it feels like a “Eureka” moment every time a robot figures out how to interact on its own.
Also, the people I get to work with: my colleagues and my students. It’s such a privilege to collaborate with some of the brightest minds in AI!
And finally, the students I get to teach. Berkeley classes are huge, but that also means that you get to be the person that exposes so many students to what AI is. In particular, when I teach Intro to AI, I try to present it in a human-centered way from the start, so that our future generations learn how to think about the interaction problem as a core aspect of AI.
Which emerging or future technologies are you excited about?
I look forward to self driving cars that can coordinate with pedestrians and other drivers, and that feel safe and comfortable, enabling their passengers to anticipate their behavior and be at ease. This is the most immediate of the applications we work on, because there has already been so much progress on the functional aspects of perceiving the world, planning safe trajectories, etc. But still, these cars tend to treat people as obstacles that they must avoid collisions with, not really as other agents that are trying to coordinate with them. It’s exciting to tackle this interaction-centric aspect of driving!
I’m also looking forward to techniques for tackling the safety and transparency aspects of AI systems. All AI agents make decisions to optimize some objective function, and it turns out it is really hard for us to specify good objective functions. If we mis-specify them, then we end up with unintended consequences that make us react with a “wait, that’s not what I meant!”. A classic example is in the Russell and Norvig book, where a vacuuming robot is rewarded for sucking dust in, and figures out that it can dump the dust out repeatedly and suck it back in. Oups, that is not what we meant when we wrote down that objective. So there is really a need for techniques that can work together with people to help arrive at the correct objective function to optimize. This is a core part of our mission in the Center for Human-Compatible AI at Berkeley.
Do you feel you have faced challenges in your field because of your gender?
I think I’ve had both challenges and opportunities. The challenges stem from the implicit biases that we, humans, sometimes have, many of them based on gender. I’ve been advised, for instance, to add more equations in my slides that my male counterparts do because it’s easier for people to, without even thinking about it, pigeonhole me as having relative less technical depth. I’ve also been advised to not raise my pitch at the end of sentences, because it makes me sound less confident. That said, I think it’s much easier for people to notice you when you are female in a male-dominated field, and that certainly has advantages in terms of the visibility of my research.
What advice would you give to someone starting a career in technology?
To take math seriously. It’s useful tools, it’s a way of thinking, and it’s the ability to pick up new tools more quickly, so it’s not to be underestimated. Beyond that, to compensate in ambition for what they lack in skill: the most successful people I know are definitely smart, but above all they are driven, putting in a lot of hard work and going above and beyond what they are asked.
See our Women in Tech & Science series for more Q&As.
Learn more about the advancements of intelligent machines at the Machine Intelligence Summit, taking place alongside the Machine Intelligence in Autonomous Vehicles Summit in Amsterdam on 28–29 June. Meet with and learn from leading experts about how AI will impact transport, manufacturing, healthcare, retail and more. Tickets are now limited, register to attend here.
Early Bird passes expire on 12 May. Register to attend now.
Opinions expressed in this interview may not represent the views of RE•WORK. As a result some opinions may even go against the views of RE•WORK but are posted in order to encourage debate and well-rounded knowledge sharing, and to allow alternate views to be presented to our community.