Robotics and AI: The future is in Rosman’s lab

There is no end to Professor Benjamin Rosman’s passion and enthusiasm for all things artificial intelligence and robotics.

In his upstairs laboratory at the University of the Witwatersrand, it’s as if the future comes to life before your eyes.

Rosman, affectionately known as Benjy, is a professor in the School of Computer Science and Applied Mathematics at Wits, where he runs the robotics, autonomous intelligence and learning laboratory.

Robotic dogs, an intelligent arm that can play chess, a spider-looking robot that was 3D printed in its entirety, are only a part of the menagerie. It’s a playground of the future, but we have the minds and the technology to realise all of this at home, right now, he says.

ALSO READ: Google reveals South Africans searching for AI at record numbers

Benji, as he prefers being called, is the kind uber-super cool guy everyone wants to know. He speaks around the world on topics ranging from artificial intelligence through to robotics and is widely considered one of the global experts in the field.

Artificial intelligence

He says: “I studied computer science because I thought this was the coolest thing I could do. And then somewhere along the line I discovered artificial intelligence and was instantly hooked.”

Rosman has been working in the field of robotics and AI for over a decade, and his research focuses on developing intelligent machines that can operate autonomously in complex environments.

He focuses on reinforcement learning and decision-making in autonomous systems or artificial intelligence-driven machines, and how learning can be accelerated through knowledge gained from solving problems as well as the role that Africa can play in shaping the future of these technologies.

Potential of technology

When he speaks about artificial intelligence and the nature of technology, and the fact that so much has so quickly become available to people, he lights up at the potential it holds.

Said the professor: “In our lab, we do a lot of primarily theoretical work, and we believe strongly in driving the science.

“We’re working on robotics a lot and the kinds of areas that could be quite impactful around things like controlling robots in a warehouse to be able to really optimise the way you get things delivered”.

ALSO READ: AI in South Africa carries the usual risk, plus unique dilemmas

But there’s more… He shared: “You might have a question of how to optimally grow a crop that can give the maximum yield and this might involve deciding when to water it, how to change the soil or the fertilizer.

“We’ve been trying to look at these kinds of problems as applications of the theoretical work that we’ve been doing.”

Rosman highlighted the advances in machine learning and the ability to create more sophisticated algorithms, the recipe that computers use to fulfil tasks or to perform complex calculations and actions. He added that this has led to significant progress in areas such as computer vision, natural language processing, and speech recognition.

ChatGPT, Google’s Bard and Microsoft’s Bing-ChatGPT hybrid search engine are all examples of natural language interfaces that allow people to communicate with the vast knowledge of the internet’s collective deposits of, well, human knowledge.

Computers are becoming more mobile and are capable of performing more complex tasks, he says.

Advances in robotics

Rosman drew on the near past and said that robots are not what they used to be. The advances have been spectacular.

He said: “Ten years ago, you’d have these clunky things, that were the butt end of jokes, about the robot dance. But now there’s some very fluid motion. They’re able to jump and balance and all sorts of things. And now there’s text technologies, at the same time as self-driving cars and at the same time AI is getting better at medical diagnosis. I think it’s just fascinating that so much is happening in every direction.”

But while robots are becoming more intelligent, they still lack the ability to understand human language and gestures, which makes it difficult to communicate with them effectively. They also need to be able to operate autonomously in unpredictable and unstructured environments.

Rosman noted that robots are good at performing repetitive and predictable tasks, but they still struggle with tasks that require a high degree of flexibility and adaptability. But he sees a bright future with almost unlimited applications that lie untapped – technologies to solve some of the world’s most pressing problems.

Robots could for instance be used to perform tasks in dangerous or hazardous environments, such as mining, firefighting, and disaster response. Robots could improve healthcare, by performing tasks such as medical imaging and diagnosis.

NOW READ: PODCAST | Tech Check: Exploring the future with CTRL Robotics