This blog post was guest-written by Nora Ayanian, Assistant Professor 

Diversity is crucial when solving challenging problems, whether you work with people, devices, or robots. For more than a decade, I’ve worked in multi-robot coordination, studying how to make teams of robots work better together. And in that time, we’ve made a lot of progress in putting robots to work.

You might not realize the huge impact robots have had on our lives: they make our cars and airplanes stronger and safer, our mobile devices smaller, and even get our packages to us faster. All of these things happen behind the scenes, in buildings with lots of instrumentation and infrastructure, including high-bandwidth communications, cameras, and other sensors. The reason robots have thrived in manufacturing and warehousing is because they have this infrastructure, and the environments are predictable and highly controlled.

But many of society’s toughest problems are in unstructured environments, environments that are not instrumented, controlled, or even well-known or understood. Robots’ potential to operate in these dangerous conditions makes them excellent candidates for these tasks. Imagine teams of robots searching for survivors and delivering food and supplies to hard-to-reach areas after a natural disaster like a tsunami or earthquake, exploring dangerous areas with remote sensing to find new sources of clean water, or patrolling waterways to prevent the dumping of toxic waste.

While robots have been deployed in such situations before, they’ve always been remote-controlled by expert operators. However, in many scenarios, waiting for experts to arrive wastes valuable time, and can mean all the difference between life or death for those in need. The operators themselves also have limited situational awareness, meaning it can be hard for them to understand what is going on around the robot.

To accomplish these tasks effectively and quickly requires autonomy, or the ability for the robots to operate and make decisions on their own. Without the infrastructure we typically lean on to ensure the robots operate both effectively and safely, though, we need new solutions that will be successful in these tough environments. I believe what we need to get these solutions is diversity, both in the teams that develop these solutions and in the solutions themselves.

Let’s start with diversity in the teams working on these problems. Research has shown that diversity in skills within a workgroup leads to higher-quality solutions. Add to that different life experiences, different perspectives, and the ability to discuss ideas respectfully, and you have a highly functional workgroup. My research group focuses on solving hard problems, so I aim to have a diverse group of people that can bring their different skills and experiences to the table. Without these valuable perspectives, I don’t believe we can tackle these difficult societal problems.

But I realized a couple years ago that I wasn’t approaching my research problems the same way. While I was focusing on diversity within my research group, I wasn’t thinking about using diverse approaches to solve multi-robot problems. This inspired me to rethink how we solve multi-robot problems.

The way we solve multi-robot problems right now is to uniformly apply one control policy to all of the identical robots in the team. For example, imagine we’re trying to monitor air quality with a team of physically identical aerial robots. If we considered all the factors that could affect the problem, the robots, and their capabilities, we might have too many factors to consider and our problem would be intractable.

Currently, we make simplifying assumptions to make the problem solvable, like estimating variable wind speed or potential changes in temperature. Then we construct the best solution for that problem and apply it to all of the robots. But when we deploy the robots, the environment isn’t exactly the same, the temperature is higher and the wind is stronger; our optimal solution across uniform aerial robots is no longer optimal! In fact, it might not work at all!

Wouldn’t it be better if the team of robots each had a slightly different control policy, created with different assumptions? In contrast to the uniform approach, some of the robots will work better than others, which means the robots could learn from one another. Which strategies worked and which didn’t? How can each adjust to perform its own task? By using a diverse approach, we can leverage the strengths of the different control policies under different conditions.

Beyond diversity in control, we can also consider diversity in the physical capabilities of robots. Imagine that same team of aerial robots assisted by robots on the ground. The robots on the ground could provide additional information such as temperature, position, topography, and satellite communications via hardware the aerial robots might not be able to carry. They could also perform computation, telling the aerial robots where to go and mapping the air quality, allowing the aerial robots to use more of their on-board energy for sensing.

Diversity is the key to using robots to solve some of society’s most pressing challenges. Diversity in robot capabilities and control, as well as diversity in the people that develop these hardware and software solutions, will ensure that we make the largest impact possible.

Getting involved in IT is an opportunity to confront these global challenges with your unique perspective. We need more people of all backgrounds involved in technology, especially women! Join me as we jump into the tech revolution on November 29 as part of the Women Rock-IT Cisco TV series.

Register today to take part in my talk during the session: “


Austin Belisle

No Longer with Cisco