Driverless Multipurpose Vehicles for Sustainable Urban Road Transportation

QC 260529

Abstract

A driverless multipurpose vehicle (DMV) is an autonomous road vehicle designed to execute multiple transportation functions, such as freight distribution and passenger transportation, within daily operations. They may reduce fleet size, energy consumption, and operating costs in urban transportation, yet their system-level implications remain poorly understood. Existing fleet optimization models rarely integrate realistic energy estimation based on key vehicle-level and transportation system-level factors on real urban road networks, limiting the evidence base for evaluating DMV deployment.

This thesis assesses the energy and operational implications of DMV deployment in urban road transportation from a system-level, operational perspective. It addresses three research questions. First, how the energy consumption of DMV fleets can be estimated and compared against human-driven battery-electric vehicle (BEV) and combustion vehicle (CV) fleets under realistic urban operating conditions. Second, how key vehicle-level and transportation system-level factors can be integrated into fleet-level optimization on realistic urban road networks. Third, what energy and operational implications emerge from DMVs with an interior-reconfigurable architecture (IRA) type compared with human-driven BEV and CV fleets across multiple cities and operational scenarios. The thesis also proposes a practice-based taxonomy of eight architectural strategies of DMVs grounded in design for changeability theory.

The methodological contribution is a two-stage optimization framework coupling fleet sizing, mix, and routing with a deterministic microscopic energy consumption model on real urban road networks. The first stage solves energy-minimal shortest path problems incorporating edge-specific driving profiles and key vehicle-level and transportation-system-level factors, producing a reduced graph. The second stage solves the novel Fleet Size and Mix Electric Vehicle Routing Problem with Simultaneous Pickup and Delivery on this reduced graph. The energy consumption model is evaluated against measured energy data from 18 trips of a battery-electric truck operating on a fixed urban route in Östersund, Sweden: it overestimates absolute energy use but reproduces subroute energy rankings, supporting its use for comparative fleet assessment.

The framework is applied across Stockholm, Paris, and Lisbon with 50 transportation operations per city and per fleet type, and four objectives. CV fleets consistently have the highest energy consumption, while BEV and DMV fleets show comparable energy use across all cities and objectives. Under cost minimization, DMV fleets achieve the lowest total cost (median approximately €430 per day) compared to CV (€450) and BEV (€730) fleets, driven primarily by the elimination of driver labor. Separate analyses show that business-as-usual operations with dedicated freight and waste fleets consume approximately 50–80% more energy than corresponding simultaneous operations, indicating substantial multipurpose capability benefits. However, because all fleet types perform the same combined service in the main comparative study in the thesis, the results primarily reflect the effect of automation rather than the full multipurpose capability of DMV fleets.

The modeling framework supports energy-informed decisions from strategic fleet procurement and technology adoption, through tactical fleet sizing and deployment, to operational routing. It provides a foundation for future research on larger problem instances requiring approximate solution methods, co-modal freight and passenger operations, additional DMV architectural strategies, and formulations that account for resilience and accessibility.

Link to DiVA