Callum Barnes
About
Callum Barnes has a First class Physics BSc from the University of Kent and was awarded the Physics prize in 2021 for achieving the highest grade in the cohort. He has experience using Machine Learning techniques with Santander UK to predict the economic performance of businesses alongside applying Machine Learning techniques on urine sample data to detect doping agents. He is one of the founding members of the Division of Natural Sciences Machine Learning Network and currently is supervised by Dr Stuart Gibson. With his PhD work he has an industrial partnership with a start-up called Body Rocket (https://www.bodyrocket.cc/) with the overarching research question to investigate the application of machine learning techniques in cycling.
Industrial Partner
Body Rocket are a startup firm who have developed the world’s first on-bike system that can directly measure the drag force of a rider. This is achieved through the use of four sensors on the contact points of the bike; the handlebar, saddle and pedals. With this drag force data the coefficient of aerodynamic drag can be calculated. The aim of Callum’s research is to delve into the rich datasets of this system and through the use of Machine Learning and Artificial Intelligence techniques lower the boundary to aerodynamic insight and push the limits of cycling. Within Body Rocket, Callum worked as an intern over the summer of 2022 and has been appointed the title Head of AI in Aerodynamics.
Research interests
In the first year of Callum's PhD he investigated the impact of shuffling on the saddle. Shuffling is a well-known phenomenon within the cycling fraternity and is the action of sliding forward onto the tip of the saddle, when aggressively cycling, then sliding back. Quantifying this can be accomplished with the Body Rocket system and Callum is one of the only researchers to have investigated this problem. This research is currently on-going and he has presented this research at the International Sports Engineering Seminar 2023 (where he won the prize for best student presentation) and at the Science and Cycling conference 2023 in Bilbao at the start of the Tour De France.
Within his first study, Callum designed and built a unique piece of optical equipment to determine the position of a rider on the saddle. This involved the creation of a camera with an infrared flashing to flash on retro reflective markers placed on the back of the rider. Through the use of bespoke computer vision code the size of these markers are identified and distance from the camera determined. This acted as a validation of the Body Rocket system to determine the position of someone on the saddle.
Moving forward into the second year of his PhD. Callum’s research includes the application of Machine Learning techniques on Body Rocket system data to determine the position a rider is in on the bike. Using a variety of techniques this study is also on-going and an abstract has been accepted to be presented at the Science and Cycling conference in Florence 2024. Additional work in his second year includes investigating the application of computer vision techniques to determine the joint location of riders on a bike with the aim to tie this into the Body Rocket system data.
With recent advancements in Large Language Models (LLMs) and Generative Pretrained Transformers (GPT's) from the likes of OpenAI, Meta and Google, Callum is taking advantage of these technologies with full force. The aim of this, perhaps most exciting, area of research is to create an artificial intelligence chatbot that can interpret the proprietary Body Rocket data and guide a rider through the process of getting more aerodynamic. With the Body Rocket system as a drag measurement device it already lowers the boundary into learning about how you interact with the air, however, just looking at drag data and trying to interpret it can be extremely daunting. The aim of this chatbot is to interpret this data, learn an individual, and tell them how to get faster, with no guessing, no false information, just facts, knowledge and understanding tailored to you.
Supervision
Supervised by Dr Stuart Gibson.
Professional
https://www.jsc-journal.com/index.php/JSC/article/view/822
(C. Barnes et al., 2024) in prep