Investigation of User Behavior in Pedal-Assisted Vehicles: From Field Testing to Driving Cycle

In recent years, electric cargo (e-cargo) bikes have been increasingly adopted as a sustainable alternative for urban logistics and last-mile delivery, particularly in densely populated areas where traditional vehicles face traffic congestion and access limitations. This study aims to develop a representative driving cycle for e-cargo bikes based on real-world cycling data. An instrumented Long John-type e-cargo bike was used to collect naturalistic data from four different riders covering a total of 50 km along a predefined route in the city center of Florence, selected in collaboration with the Italian postal service provider (i.e., Poste Italiane) to reflect typical delivery operations. The driving cycle was generated using a Markov chain Monte Carlo (MCMC) method, modeling the stochastic transitions of vehicle speed and acceleration values. The resulting driving cycle, defined as the Florence cargo bike driving cycle (FCBDC), achieved an error of 2.1% on the Speed Acceleration Probability Distribution (SAPD) root sum square difference; although minor losses in peak acceleration values were observed due to data smoothing and discretization, the synthesized driving cycle effectively reproduces the dynamic characteristics of e-cargo bike riding. While the study is limited to a single route and is equivalent to simulated postman behavior, it provides valuable insights to guide the future development and optimization of e-cargo bikes for sustainable mobility operations.