At the time, my BSc degree in Mechanical Engineering from UKZN was the most significant milestone in my life,’ exclaims Mr Mohale Molefe. ‘You can only imagine what I must be feeling now to be graduating with a master’s degree. My family is so excited.’
‘Studying for my Master’s degree in Computer Engineering at UKZN has been an exciting and nerve-wrecking experience,’ he added, ‘but with support, advice and the hands-on approach of my supervisor, Professor Jules Raymond Tapamo, changing from one engineering discipline to the other felt seamless and manageable.’
Completing an undergraduate degree in a different discipline to his master’s degree has made Molefe’s skills set more varied and diverse. ‘Modules such as Programming, Robotics, Mechatronics and Control Systems made it easy for me to continue my postgraduate studies in Computer Engineering as I was already familiar with some of the modules and concepts,’ he said.
His multidisciplinary knowledge also opened up employment opportunities. Molefe is employed by Transnet Freight Rail, where his main focus is to introduce Artificial Intelligence techniques for safe and reliable transportation of freight.
Molefe’s MSc research applied image processing and machine learning techniques for automated detection and classification of rail welding defects. Conventionally, railway industries have used radiography testing methods to inspect possible defects that could have occurred during the welding procedure. However, this is a long, costly and subjective process as it relies on human expertise. Molefe’s research looked at how the principles of image processing and machine learning algorithms can be utilised to automate such investigations.
‘Throughout history, railway transportation has played a vital role in transporting heavy freight and passengers at a lower cost than other transportation modes,’ explained Molefe. ‘Furthermore, it plays a significant role in developing the country’s economy and enabling the growth of other sectors such as mining and agriculture.
‘Therefore, it remains a crucial task for all railway maintenance personnel to ensure safe and reliable train transportation. Railroad failures such as rail breaks are directly linked to train derailments and cause the loss of lives and revenue. Most rail breaks have been traced to welding defects which were not detected owing to human error.’
Molefe’s study proposes an automated method that enables rail welding defects to be investigated in a fast, reliable and objective manner.
Molefe is currently pursuing a PhD in Graph Theory and Graph Representation Learning at UKZN under the supervision of Tapamo. This is a new and emerging field within the domain of Machine Learning. His study aims to design a model to classify different road types for realistic cities from the road network graph dataset. This would assist road users in planning their road trips by avoiding congested routes and routes with many intersections. Such a model can also be integrated into interactive maps to provide road users with useful traffic information.
The world really is an oyster for this old school rap music lover.
Words: Swastika Maney
Photograph: Supplied
