Machine learning for the activation of contraflows during hurricane evacuation

Contraflows are a critical part of an emergency evacuation plan. In most cases, a contraflow lane reversal will double the capacity of key evacuation routes. The Contraflow plan for the evacuation of southeast Louisiana during a hurricane threat uses a typical schedule for the activation of contraflows based on the predicted time of landfall. This work will apply machine learning techniques using real-time traffic data to schedule the activation of contraflows. Optimizing the Contraflow plan should increase the effectiveness of the evacuation plan by increasing the flow of evacuation traffic based on demand and retaining the availability of incoming traffic until contraflow lanes are needed. These techniques could be applied to other locations, including those without an existing evacuation plan.