Semi-synthetic Data and Testbed for Long-Distance E-Vehicle Routing

Barauskas, Andrius; Brilingaitė, Agnė; Bukauskas, Linas; Čeikutė, Vaida; Čivilis, Alminas; Šaltenis, Simonas · 2021 · Crossref

DOI: 10.1007/978-3-030-85082-1_6

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the lack of realistic, large-scale datasets required for robustly testing advanced routing algorithms for electric vehicles (EVs). As autonomous and electric mobility grows, routing algorithms must handle complex, multi-objective optimizations involving time-dependent traffic, energy consumption, and charging infrastructure availability. Existing benchmarks often require extensive manual data preparation or rely on unrealistic traffic models. To solve this, the authors present a modular testbed architecture and a semi-synthetic data generator that produces time-dependent travel-time and energy-use weights for road-network graphs, capturing inherent prediction uncertainties. The testbed integrates several open-source tools and data sources to generate realistic semi-synthetic data. Road network data is sourced from OpenStreetMap, filtered for vehicle roads, and converted into a routable directed graph. Traffic simulation is performed using the SUMO simulator, calibrated against TomTom’s Traffic Index and Google Maps data to ensure realistic congestion and free-flow speeds. Energy consumption is estimated using a Vehicle Energy Model that incorporates EV-specific parameters (e.g., mass, drag coefficient, battery capacity) and road-dependent factors like elevation and slope, derived from CGIAR-CSI SRTM data. Charging station data, including availability profiles and charging power curves, is sourced from Open Charge Map and other public datasets. The system provides an API with functions to calculate paths, estimate energy and time intervals, and determine charging and waiting times, modeling uncertainty by providing interval estimates rather than single-point predictions. Experimental evaluation was conducted using the KaTCH routing engine on a dataset representing Germany. The authors tested 16 long-distance trips (8 round trips) with departures at non-congestion (00:00) and congestion (16:00) times. Results were compared against commercial data from TomTom. The testbed produced travel-time and energy-consumption estimates closely aligned with commercial services, though it was slightly more conservative during non-congestion hours and more optimistic during congestion hours. Uncertainty intervals were generated, with lengths reaching approximately half an hour for long trips. Scalability tests demonstrated that routing query costs remained nearly constant regardless of trip length, while energy computation costs grew linearly with path length. The system successfully handled large-scale data, utilizing approximately 21 GB of RAM for the routing structure. The significance of this work lies in providing the first comprehensive testbed for experimenting with advanced EV routing algorithms. By automating the generation of realistic, time-dependent, and EV-specific data, the testbed reduces the preparation burden for researchers and enables more rigorous evaluation of heuristic routing algorithms. The ability to model uncertainty and integrate diverse data sources (traffic, elevation, charging infrastructure) makes it a valuable tool for developing efficient routing solutions for autonomous electric fleets. The authors suggest future extensions could include life-cycle profiles and more flexible experimental setups.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• exposure measurement