Transformer Networks for Predictive Group Elevator Control

  title={Transformer Networks for Predictive Group Elevator Control},
  author={Jing Zhang and Athanasios Tsiligkaridis and Hiroshi Taguchi and Arvind U. Raghunathan and Daniel Nikovski},
  journal={2022 European Control Conference (ECC)},
We propose a Predictive Group Elevator Scheduler by using predictive information of passengers arrivals from a Transformer based destination predictor and a linear regression model that predicts remaining time to destinations. Through extensive empirical evaluation, we find that the savings of Average Waiting Time (AWT) could be as high as above 50% for light arrival streams and around 15% for medium arrival streams in afternoon down-peak traffic regimes. Such results can be obtained after… 

Figures from this paper



Marginalizing Out Future Passengers in Group Elevator Control

A probabilistic model of how these arrivals affect the behavior of elevator cars at the lobby is developed, and it is demonstrated how this model can be used to very significantly reduce the average waiting time of all passengers.

Optimal dispatching control for elevator systems during uppeak traffic

An optimal dispatching controller for elevator systems during uppeak traffic is developed using a Markov decision problem formulation with a batch service queueing model consisting of a single queue served by multiple finite-capacity bulk servers.

Group Elevator Scheduling With Advance Information for Normal and Emergency Modes

Numerical testing results demonstrate near-optimal solution quality, computational efficiency, the value of future traffic information, and the potential of using elevators for emergency evacuation.

Personalized Destination Prediction Using Transformers in a Contextless Data Setting

This work proposes a Transformer model to predict destinations from partial trajectories and demonstrates its use on two datasets from different domains, including a simulated indoor dataset and an outdoor taxi trajectory dataset.

Establishing the upper performance limit of destination elevator group control using idealised optimal benchmarks

The concept of an idealised optimal benchmark (IOB) is used in many engineering disciplines. An example of an IOB from the area of thermodynamics is the formula for evaluating the maximum possible

Sensor-aware elevator scheduling for smart building environments

Elevator Traffic Handbook: Theory and Practice

This chapter discusses the principles of classical lift traffic design and the role of uppeak traffic calculations in this design process.

Attention is All you Need

A new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely is proposed, which generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.

Tracking people in mixed modality systems

This paper describes a method that allows to evaluate when the tracking system is unreliable and present the data to a human operator for disambiguation, and presents an approach to tracking in mixed modality systems, with a variety of sensors.

The smart floor: a mechanism for natural user identification and tracking

A system for identifying people based on their footstep force profiles and its accuracy against a large pool of footstep data is tested and it is shown that the effect of footwear is negligible on recognition accuracy.