OmniOpti is the company that is performing optimization of delivery vehicles routes and they wanted to examine the possibility of applying HPC to optimize the route of delivery vehicles and examine the cost-effectiveness of the whole process. Our solution adds another dimension to the route optimization problem - using more and better alternative routes. This provides the users with more choices to choose from and consequently achieve big savings.
Route optimization is a process with high computational costs, so it is a perfect match for HPC. Higher computer power leads to less costs.
We can optimize all kinds of distribution operations, mobility services and other logistic operations, but currently focusing on road transport.
Apart from decreasing fuel usage we may also reduce congestions. Our solution can be used for improving centralized traffic management solutions, which will be especially useful for autonomous vehicles. This centralized traffic management system is not yet an off-the-shelf solution, but needs some additional research and development, possibly with several partners.
University’s goal was to investigate the possibility of HPC usage for route optimization problem solving. Our goal was to investigate the possibility of HPC usage for route optimization problem solving. The chosen approach is to convert existing route optimization algorithm, written sequentially in Java, to parallel Python code. For that, existing solutions were chosen.
For achieving the goals defined above, their team has spent approximately 20 man-days on research and 20 man-days on implementation of described solutions. Approximately 75% of the total problem is solved.
After HPC testing is performed, OmniOpti will need to evaluate profitability of HPC usage for route optimization problem solving.