"We believe our success originates from a deep understanding of the data, following the principle of simplicity* and taking advantage of the concept wisdom of the crowd**, says Peyman Sheikholharam Mashhadi, postdoctoral researcher at Center for Applied Intelligent Systems Research (CAISR) and one of the team members in the Aramis challenge.
The objective of the Aramis challenge was to build AI models (specifically machine learning models) to accurately detect faults and predict failures in industrial equipment that is operating under a constantly changing and evolving environment, for example heavy-duty vehicles for transportation tasks, like electric buses and trucks. The team built these AI models based on a large dataset, containing sensor data from industrial equipment, given by the Aramis challenge organisors.
Using AI to predict failure and plan maintenance
In reality, for example when applied to a fleet of city busses, this sort of AI system can be used to predict machine failure and plan the maintenance accordingly.
"The beauty in what we did from my point of view is that we applied the concept wisdom of the crowd to the way we operated as a team. At first, each team member worked on resolving the challenge individually. Only at a later stage, we combined our efforts and created an ensemble solution that consume our individual solutions predictions as an input. The resulting ‘team predictor’ proved to be more capable than any individual solution", says Mohammed Ghaith Altarabichi, PhD student at CAISR.