A dispatch algorithm using trained machine learning models to improve elevator assignment decisions over time based on actual traffic data.
Full Definition
Machine learning dispatch is the application of supervised or reinforcement learning models to the elevator group controller dispatch problem. The system trains on historical traffic data (time of day, day of week, floor demand patterns, car loading) and continuously refines its dispatching strategy to minimise average waiting time and maximum waiting time across the building. Unlike static dispatch rules, ML models adapt as the building's occupancy patterns change (new tenants, seasonal patterns, special events). Models are typically hosted on edge controllers or in the cloud and are updated through regular retraining cycles. Integration with access control and visitor management systems further enriches the demand signal available to the model.
