William Hugh Burrough Smith

William Hugh Burrough Smith (PhD 2017–2022) trained robots to navigate autonomously using machine learning, developing methods to recognize places under changing conditions and detect scenes of interest. He is now a Machine Learning Research Engineer, creating advanced AI solutions for real-world applications.

PhD start and end years	2017 - 2022

What was your PhD research about?

My PhD was about training robots to navigate autonomously with machine learning. I used machine learning to recognise places despite visual changes due to weather and season. I also worked on differentiating specific scenes of interest such as pedestrian crossings and roundabouts from general driving scenes while navigating as well as identifying when the robot had left a pre-defined map made up of images.

What motivated you to undertake doctoral study?

I was really interested in learning more about robotics and machine learning. Rather than replicating humanity I was interested in investigating and understanding different types of intelligence that might emerge from machine learning. I believe this can give us really interesting and diverse perspectives on the world around us.

What was a highlight (one or several) of your time as a doctoral researcher?

Spending 6 months at a big audio company in Denmark on an internship there, I got to meet other students from around the world and live in a different culture. I also got a paper published at a top conference called ICRA before the end of my PhD, this represented a contribution towards the research community that I still admire deeply to this day.

What challenges did you face, and how did you overcome them?

I struggled with my intiial PhD supervisor and had to find an alternative one. I was on a government funded PhD program and was not attached to any particular project so I had a lot of freedom, but nobody was specifically invested in my project from the beginning. It was easy to feel isolated and directionless. I overcame these challenges by reaching out to a lot of different academics and putting time into explaining my work to them and taking feedback onboard, I also followed my own interests and did not try to force myself to do work I wasn't interested in.

What are you doing now career-wise, and how did your PhD prepare you for it?

Now I am a machine learning research engineer. My role is directly realted to my PhD: I get given a desired product feature and I autonomously research the target area, implement a state-of-the-art solution and potentially further improve it where time and funding allows.

What’s one key skill or mindset you developed during your PhD that you still rely on today?

Drawing conclusions based on evidence and not advocating for particular solutions, instead focusing more on the problems and appreciating their complexity. Also using scientific methods to thoroughly test and evaluate approaches to identify where they fail and therefore how they can be improved, carefully designed demos are often misleading!

What advice would you give to someone considering a PhD in Informatics?

Focus on learning the research process: identifying and understanding the state-of-the-art, implementing it, breaking it and then improving it. Don't get too attached to ideas that don't work out, it's inevitable that you are going to be wrong and it's okay to move on. Follow your own curiosity and ask question with experiments, this will give you key insights into your field that make up a PhD. Try to find a supervisor that supports you, take their advice and explain your work as clearly as possible, nobody apart from you really understand it!