Towards an Improved Time Efficiency for Elevator Calls using HOG+SVM to Handle Phantom Floor Calls

Elevator time efficiency is a crucial factor in determining the workflow of a building. Good elevator dispatching and traffic management systems will provide minimum travel and waiting times, and an overall better user experience. One of the issues hindering the efficiency of elevators especially in high-rise buildings is the often occurring phantom floor calls. Phantom floor call is a condition where an elevator car stops at an empty floor with no passengers waiting to go in or out of the elevator. This paper proposes a method to minimize the occurrence of phantom floor calls by integrating HOG+SVM human detection algorithm in surveillance cameras placed in elevator hallways. The algorithm would assist elevators in skipping floor calls based on the detected human presence from the floor’s surveillance camera image. The experiment was conducted by running two elevator simulations simultaneously, one with integrated human detection, and one without. Our method considerably reduced phantom floor calls and our findings indicated that overall elevator stops were reduced by an average of 63.66%, resulting in a more time-efficient elevator system. We also found that our method scaled well with varied building heights and floor number counts.