Fleet operations, curb usage, and parking search for shared autonomous vehicle (SAV) fleets

Access full-text files

Date

2021-12-03

Authors

Hunter, Christian Bryan

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Advances in communication, information technologies and automation in vehicles have birthed new transportation services, incuding shared autonomous vehicles (SAVs). SAVs are on-demand self-driving taxis, with flexible routes and schedules, able to replace personal vehicles for many trip types in the near future. The siting and density of pick-up and drop-off (PUDO) points for SAVs, much like bus stops in busy downtowns, can be key in planning SAV fleet operations, since PUDOs impact SAV demand, route choices, passenger wait times, and network congestion. Unlike traditional human-driven taxis and ride-hailing vehicles like Lyft and Uber, SAVs are unlikely to engage in quasi-legal procedures, like double parking or fire hydrant pick-ups. In congested settings, like central business districts or airport curbs, SAVs and others will not be allowed to pick up and drop off passengers wherever they like. Recognizing such situations, this thesis models the impact of different PUDO locations and densities as well as off-street parking provision on SAV fleet operations across the Austin, Texas metro area. This work designs and implements new code for use in POLARIS, an agent-based simulation platform originally developed by researchers at Argonne National Laboratory. Simulation results for 18 distinct scenarios suggest that the ideal SAV fleet to serve the 80 core square-miles of Austin has 4000 SAVs, charges a $1 fare per mile, and has PUDOs placed every three city blocks in the central business district (CBD). This scenario reduces the total number of curb spaces required for PUDOs by 54% and lowers SAV ridership by 6% relative to the scenario with an identical SAV fleet and fare but PUDOs on every city block in the CBD. While each individual PUDO may have more spaces when they are more spread out, aggregating PUDOs into few locations reduces the total number of spaces required. The POLARIS program was expanded to force SAV to find available on- or off-street parking after completing trips, instead of assuming that there is available curb parking at the destination and simply idling in place (which other SAV simulations do). This includes finding the least expensive parking option recognizing travel distance to and cost of parking at free and paid on-street locations as well as some paid public garages and lots. The code now maintains a database of parking sites and available spaces every 6 seconds. The SAV parking code has been tested on the 79 square-mile Bloomington, Illinois network with 750 SAVs and 56 passenger trips per SAV per day. Policies that do not permit curbside idling of SAVs at passenger pickup and drop-off locations (trip end addresses) result in 13% higher empty vehicle-miles traveled (eVMT), while Bloomington response times to pick up SAV users fell by 9.5%, from 4.22 minutes on average to just 3.82 minutes. This suggests that routing SAVs to designated sites after they are done carrying passengers can place them closer to the start of new trips. Because the parking locations were chosen randomly for the Bloomington study, the response time may respond differently of vehicles are sent towards areas of expected demand instead of the closest or least expensive location. Further parking analysis on larger networks, like the 6-county Austin region used for PUDO modeling, will improve on these initial findings.

Description

LCSH Subject Headings

Citation