Traffic jams are a particularly difficult global problem. They cause considerable economic losses. These losses are particularly significant for large cities in developing countries such as Dakar. Hence the need to optimize urban traffic by studying the impact of the various factors determining traffic flow. Traffic simulators are an inexpensive way of studying and improving traffic flow. However, most existing simulators are adapted to the traffic context of cities in developed countries, for which they were primarily created. Their use in African cities such as Dakar would be ineffective if they were not adapted to the local context, where pedestrian crosswalks are almost non-existent. In this work, we propose a pedestrian decision model for crossing the road in illegal locations. We demonstrate its effectiveness through three simulation scenarios using SUMO, an open-source, microscopic, and very popular traffic simulation package designed to handle large networks.