Akua Dickson researches safe autonomy, a future where robots help, not harm

By Chloe Cramutola

When Akua Dickson studied electrical engineering at home in Ghana, her goal was to apply mathematics to solve real-world problems that could have an impact.

In Ghana, a prominent issue is energy efficiency, she said.

“We have a lot of solar energy because we don’t have too many energy resources from other plants, so we try a lot to conserve,” said Dickson, a Systems Engineering PhD candidate in her third year at Boston University. “So at the time, I always wanted to be the kind of engineer that seeks to apply my skills to the problems that affect my community — my immediate community.”

Most of her research was designing controllers to help regulate lighting and HVAC systems so they don’t consume too much energy while they are not in use. While Dickson researched control theory and control systems in her undergraduate years in Ghana, she always wanted to go into robotics. She later applied to more interdisciplinary PhD programs — a blend of electrical and mechanical engineering through robotics — and chose the Systems Engineering program at BU.

As she started research for her PhD, she worked with robotics controllers and motion planning and found a love for safe autonomy — robots doing autonomous tasks and helping humans solve problems in a safe way, she explained.

“One of the major concerns is that as engineers we need to look at being able to trust these systems and being able to guarantee that they’re safe enough for them to interact with human beings,” Dickson said. “Because the goal of robotics is not for them to replace human beings, but for them to work hand in hand with human beings.”

In order for robots to do that, she added, they need to be guaranteed to be safe so that they can work and coexist with people without causing damage. Dickson added that this quickly became one of her major interests: How can she deploy systems that humans can trust?

To answer this question, Dickson works on non-linear control and optimization, where her main goal is to provide mathematical guarantees that will ensure robots act the way they are supposed to, she said.

In her first year and half of her second year, Dickson worked with autonomous ground robots, which are often unmanned vehicles that can perform heavy-duty tasks.

“I’m now developing a love for soft robotics, which are inherently safe to some extent because they’re made of softer material, unlike the bigger, more rigid robots that can cause a lot of damage on impact,” Dickson said.

These soft robots can perform more delicate tasks because they do not exert as much force that can harm the target. For example, Dickson thinks about soft robots autonomously performing surgeries in a safe manner.

To define what safe autonomy actually means, Dickson explained she looks at different approaches to provide formal safety guarantees. She has worked with control barrier functions, which is a concept that enables researchers to write out equations that ensure the robot is bounded within a certain area, she explained.

To do this, Dickson decomposes, or breaks down, a huge environment into smaller chunks that help make the problem appear simpler. If she can guarantee safety in a tiny “box,” then she can use that to add 100 boxes, put them together, and extend that same approach to a bigger environment.

“Sometimes we do a pre-planning problem, where the robot knows its entire environment and knows where the obstacles are,” Dickson said. “When it sees an obstacle, then it decides like, ‘Oh, let me apply my control barrier function here to avoid this obstacle.’”

To create control barriers, Dickson added that she uses polytopes, which are collections of half-spaces that bound a certain area. While they can be physical lines in the environment, they are more often mathematically represented by a set of equations, where she can see the edges of the polytope.

If, for example, she has a pentagon and there are five different edges, then she can write out equations to describe each of the edges mathematically. Then, she can use those edges to write out the control barrier function, which would guarantee that the robot stays within the edges.

According to Dickson, safe autonomy will become even more popular in the next 10 years. She added that she feels like this field will be the next phase of the world.

“Robots are definitely not going to take over, but they’re going to become our best friends and help us accomplish tasks faster and help us be more efficient,” she said. “I feel like there’s so much that we can use them for. Imagine if we were at a point where robots could actually fight fires. They could go into these places, and human beings wouldn’t even have to go as close to the danger.”

Dickson said she is unsure if she will go into the industry or stay in academia, but she is leaning more toward the industry as of now. As a roboticist and someone passionate about safe autonomy, she said she would love a more research-focused industry job. Her next steps are to advance her research in safe autonomous environmental contact for soft robot manipulators using control barrier functions.

“Having the opportunity to be one of the scientists at the forefront of these improvements and these technologies is exciting,” Dickson said. “I feel like there are really exciting times ahead, and robotics has a lot of potential. We can see that robotics today is so different from what it was even five years ago. So just imagine what we can achieve in five years.”