Optimal ambulance positioning assignment using reinforcement learning:

This study explores the optimal allocation of ambulances in Bogotá using reinforcement learning algorithms to reduce response times to traffic accidents. Through simulations based on real data and a model of the city's road network, different configurations of dispatch points are analyzed. Three key algorithms are compared: ε-Greedy, Upper Confidence Bound (UCB) and the Gradient Bandit Algorithm (GBA). The results reveal that the ε-Greedy approach achieves optimal ambulance distribution, reducing response times to an average of 5.8 minutes. This research offers a solution to improve the efficiency of emergency services, highlighting the potential of reinforcement learning in critical logistical problems.