Project Abstract

Autonomous driving is a large area of research within machine learning that can be done in both simulated and real-world environments. Within a simulated environment an algorithm can be left to train for hours as it can reset itself when mistakes are made and doesn’t require any oversight, however in real-world environments this is not the case, so it is necessary to optimise the training process as much as possible for the best results. The aim of this project is to explore machine learning techniques to train a toy car in the real-world to drive around a track optimising the performance. This includes reducing the number of crashes and the lap time, so the car can complete laps in good time consistently. This project was done in python using tools such as OpenCV, TensorFlow, and Ultralytics YOLO (You Only Look Once) to detect and track the car around the track and train a machine learning agent. The agent is trained using reinforcement learning through TensorFlow using information gained from OpenCV and YOLO.

Keywords: Python, Machine Learning, Real time

 Conference Details

Session: A

Location: Sir Stanley Clarke Auditorium at 11:00 13:00

Markers: Matt Roach, Betsy Dayana Marcela Chaparro Rico

Course: BSc Computer Science 3yr FT

Future Plans: I’m undecided