An Example of a Dynamic CPN Model to Obtain Routes in the Presence of Obstacles Detected Using Machine Learning Techniques

Obtaining routes has been a challenge during the recent history of software systems. GPS has helped this technology to be widely used. Some challenges still remain like providing routes in the presence of obstacles. The term obstacle differs depending on the scenario. It is not the same for wheelchair users than for a group of school children. In this paper, we study how to enable a set of technologies to bridge this gap. We propose to combine a graphical formalism with machine learning. Color Petri Nets are used to model a dynamic scenario where conditions, i.e. obstacles, change at real time, while machine learning is used for obstacle detection. Several techniques can be applied to obtain routes in this context. We use a simulation technique which provides fast routes in real time. We apply this technique in the context of a school visit to the Alhambra Generalife Gardens in Granada (Spain) providing a model of the gardens’ map which evolves dynamically depending on the real time data offered by the machine learning module. This module is in charge of detecting overcrowded areas in order to avoid them. Routes are obtained by simulating several users traversing the map simultaneously and the fastest is returned. The system provided can be adapted to other user necessities and scenarios.