About me

I joined the CDT because… I want to develop revolutionary novel farming technology based on new principles of agriculture.  

My favourite thing about the CDT is… Access to agricultural robotic testing facilities. 

 My favourite thing about Lincoln/What I am most looking forward to in coming to Lincoln is… Change of pace from London to a rural countryside setting. 

The University where I am going to study my PhD is…  Lincoln 

I am looking forward to going there because… I get continued access to the Riseholme site to test my creations. 

My career goal is to… Start a world leading agricultural robotics company. 

 The areas of agri-food robotics I am most interest in are… Full stack. Farm to fork. Creating new robotic agricultural systems.  

In my spare time I enjoy… Music, computer games, fitness. 

A fun fact about me is… I have 540 degrees of rotation through my wrist due to joint hyper mobility.  

MSc Project

Autonomous plant detection and geolocation for multi-robot coordination

Accurate plant species recognition is crucial in advancing agricultural robotics, particularly in the realm of precision agriculture. This field has seen significant growth due to the increasing need for precise application of agrochemicals, which benefits both cost efficiency and environmental sustainability. Monitoring individual plants to gather data for creating digital farming maps is gaining importance, especially with the rise of regenerative agriculture practices. These practices emphasize soil health and nutrient sequestration with minimal fertilizer input.

In this project, computer vision techniques are employed to address the challenges posed by agricultural environments, which are characterized by high variability in conditions. A diverse dataset was created to train a computer vision system under realistic conditions, ensuring robustness in model performance. The project leverages artificial intelligence and machine learning to automate data processing and labelling, significantly speeding up system training and model adaptability. Transfer learning, specifically using the YOLOv8 model, is utilized to fine-tune an existing model with new data, optimizing it for the task at hand.

The demonstration involves four TurtleBot robots—two scouts and two sprayers—operating in a simulated agricultural environment. The goal is to autonomously classify and geolocate 12 different plant species, accurately mapping their positions. The existing ROS1-based codebase, developed by Helan Harman, serves as the foundation for this demonstration, with enhancements including the integration of a centralized fleet management system. This work aligns closely with the Robotics and Autonomous Systems (RAS) program, focusing on machine vision, multi-robot coordination, and advanced dataset creation techniques.

PhD Project

Title to be confirmed