Data and code for "Less is more: Lightweight reinforcement learning method for traffic signal control with less observation"

The data consists of normal driving records for dozens of private cars over several months (from June 5, 2015 to June 30, 2016), with a sampling frequency of one minute. The basic specifications of the vehicles are as follows: Roewe E50 is a pure electric vehicle, weighing 1080 kilograms. It is equipped with a 22.4 kWh battery pack, and is reported to have a driving range of 170 kilometers. The raw data has been preprocessed and denoised, resulting in a final dataset containing 10,000 trips. This dataset offers substantial potential for reuse in research and analysis focused on electric vehicle energy consumption. Researchers, engineers, and policy makers can leverage this data to understand patterns, develop optimization algorithms, and inform energy-efficient practices. The dataset adheres to all applicable legal requirements. All sensitive information has been removed, and the data has been preprocessed to ensure confidentiality. There are no known legal or ethical obstacles to its use.

The package includes the codes (Codes.rar) and simulation results (simulation results.rar) related to the manuscript 'Evaluation of platooning configurations for connected and automated vehicles at an isolated roundabout in a mixed traffic environment'. Main codes are written in Python and Matlab, and the simulation results are stored in ‘.xlsx’ format.

The dataset comes from the paper "Markov probabilistic decision making of self-driving cars in a highway with random traffic flow: a simulation study" and mainly consists of three parts:

1. The "policy_training" folder, which includes the code used for training driving strategies in the highway scenario mentioned in the paper. The main function is "main_v3.cpp," and the functionality is implemented in the file "f_define_v3.cpp." Additionally, "to_strategy.m" exports the trained strategy as a state-action table for application, and "FIGURE.me" visualizes this table.

2. The "policy_application" folder contains "test.py" as the main file, which uses "MdpModel.py" to call the state-action table from the "Strategy" folder and visualizes it through "MdpModel.py." The "Map" and "Vehicle" folders store visualization elements.

3. The third part is a video of the visualization results named "simulation_demo.avi."

Purpose — Transit electrification has emerged as an unstoppable force, driven by the considerable environmental benefits if offers. Nevertheless, the adoption of battery electric bus is still impeded by its limited flexibility. The constraint necessitates adjustments to current bus scheduling plans. To this end, this paper aspires to offer a thorough review of articles focused on battery electric bus scheduling.

Design/methodology/approach — We provide a comprehensive review of 42 papers on electric bus scheduling and related studies, with a focus on the most recent developments and trends in this research domain.