Optimal Parking Management of Connected Autonomous Vehicles: A Control-Theoretic Approach
Shi'an Wang, University of MinnesotaShow Abstract
Michael Levin, University of Minnesota
Ryan Caverly, University of Minnesota
In this paper we develop a continuous-time stochastic dynamic model for the optimal parking management of connected autonomous vehicles (CAVs) in the presence of multiple parking lots within a given area. Inspired by the well-known Lotka-Volterra equations, a mathematical model is developed to explicitly incorporate the interactions among those parking garages under consideration. Time-dependent parking space availability is considered as the system state, while the dynamic price of parking is naturally used as the control input which can be properly chosen by the parking garage operators from the admissible set. By regulating parking rates, the total demand for parking can be distributed among the set of parking lots under question. Further, we formulate an optimal control problem (called Bolza problem) with the objective of maintaining the availability (managing the demand) of each parking garage at a desired level, which could potentially reduce traffic congestion as well as fuel consumption of CAVs. Based on the necessary conditions of optimality given by Pontryagin's minimum principle, we develop a computational algorithm to address the nonlinear optimization problem and formally prove its convergence. A series of Monte Carlo simulations is conducted under various scenarios and the corresponding optimization problems are solved determining the optimal pricing policy for each parking lot. Since the stochastic dynamic model is general and the control inputs, i.e., parking rates, are easy to implement, it is believed that the procedures presented here will shed light on the parking management of CAVs in the near future.
Vehicle Routing for Shared Autonomous Electric Vehicles Considering Passengers’ Uncertain Waiting Time Tolerance and Acceptable Stopovers
Xu Ouyang, Hong Kong Polytechnic UniversityShow Abstract
Min Xu (email@example.com), Hong Kong Polytechnic University
Ting Wu, Hong Kong Polytechnic University
The ridesharing system effectively allocates limited vehicle resources by organizing passengers with similar itineraries to share seats of the same vehicle, thereby reducing vehicle miles traveled and the number of required vehicles. Shared autonomous electric vehicles (SAEVs), which apply autonomous electric vehicles to ridesharing systems, have attracted rising attention from both academia and industry, thanks to their high fleet efficiency, flexible mobility, and low energy consumption. This study investigates the vehicle routing problem of SAEVs considering passengers’ uncertain waiting time tolerance (WTT) and acceptable stopovers. Specifically, a passenger is assumed to have an uncertain WTT for vehicle pick-up delays and a pre-specified number of acceptable stopovers during the entire trip. The vehicle routes of SAEVs are optimized to serve a set of pre-known passengers so that the profit of SAEV operators are maximized. To solve this problem, we first develop a robust mixed-integer program (RMIP) that explicitly formulates the number of stopovers during passenger trips and all possible realizations of passengers’ uncertain WTT. The intractable constraints associated with the uncertain WTT in RMIP are then addressed using robust optimization techniques. Since the resultant model cannot be easily solved by available solvers, we thus further develop a formulation-based two-layer heuristic algorithm by exploiting decomposable model structures to efficiently find good-quality solutions. A case study using real-life ridesharing data of Chengdu, China demonstrates the efficacy of the proposed model and the heuristic algorithm. We also analyze the impact of stopovers and fleet size on profits and the optimal solution and report practical managerial insights.
Incentive-based Decentralized Routing for Connected and Autonomous Vehicles
Chaojie Wang (firstname.lastname@example.org), Georgia Institute of Technology (Georgia Tech)Show Abstract
Jian Wang, Ningbo University
Srinivas Peeta, Georgia Institute of Technology (Georgia Tech)
Routing strategies under the aegis of dynamic traffic assignment have been proposed in the literature to optimize system performance. However, challenges have persisted in their deployment ability and effectiveness due to inherent strong assumptions on traveler behavior and availability of network-level real-time traffic information, and the high computational burden associated with computing network-wide flows in real-time. To address these gaps, this study proposes an incentive-based decentralized routing strategy to nudge the network performance closer to the system optimum for the context where all vehicles are connected and autonomous vehicles (CAVs). The strategy consists of three stages. The first stage incorporates a local route switching dynamical system to approximate the system optimal route flow in a local area based on vehicles’ knowledge of local traffic information. The second stage optimizes the route for each CAV by considering individual heterogeneity in traveler preferences (e.g., the value of time) to maximize the utilities of all travelers in the local area. The third stage leverages an expected envy-free incentive mechanism to ensure that travelers in the local area can accept the optimal routes determined in the second stage. The study analytically discusses the convergence of the local route switching dynamical system. We also show that the proposed incentive mechanism is expected individual rational and budget-balanced, which ensure that travelers are willing to participate and guarantee the balance between payments and compensations, respectively. Further, the conditions for the expected incentive compatibility of the incentive mechanism are analyzed and proved, ensuring behavioral honesty in disclosing information.
Daily Load Planning Under Different Autonomous Truck Deployment Scenarios
Lama Al Hajj Hassan, Northwestern UniversityShow Abstract
Hani Mahmassani, Northwestern University
Mike Hewitt, Loyola University
This paper presents and tests modified service network design formulations that account for five levels of truck automation in a daily load planning setting. Given daily updates of load information, the paths for the five deployment scenarios are adjusted using two daily updating methods. Both methods start with a base plan in which paths are generated based on the historic daily distribution of load dispatches during an average week .The two methods are: (1) Option 1: re-optimization of pre-booked loads and new requests, and (2) Option 2: optimization of new requests only. The solutions of the two options are compared to the hindsight plan which assumes complete information of actual requests placed. Results show that the cost savings achieved with re-optimization (Option 1) compared to insertion (Option 2) increase with more demand variability; this outcome is consistent across all fleet mixes. When most of the loads are new arrivals, the computational time saved with insertion is less attractive than the possible cost savings achieved with re-optimization. With daily re-optimization, most of the plan changes adjust the terminals visited by the load compared to just changing the dispatch and arrival times along the load’s path.
ASSESSING LONG-TERM IMPACTS OF AUTOMATION ON FREIGHT TRANSPORT AND LOGISTICS NETWORKS BY A LARGE-SCALE LRP INTEGRATED IN MICROSCOPIC TRANSPORT SIMULATION FOR STRATEGIC TRANSPORT AND LOGISTICS NETWORK PLANNING
Elija Deineko, Deutsches Zentrum fur Luft- und Raumfahrt DLR Standort BerlinShow Abstract
Carina Thaller, Deutsches Zentrum fur Luft- und Raumfahrt DLR Standort Berlin
Gernot Liedtke, Deutsches Zentrum fur Luft- und Raumfahrt DLR Standort Berlin
Up to now, bulk transports have been carried out via a hub-and-spoke network in the general cargo sector. However, it is expected that the use of autonomous vehicles will enable a more flexible delivery. Such developments may, economically, make sense for shippers. From an ecological point of view, also negative effects can be expected due to enhanced transport perfomance. In the framework of this research, we investigate the impacts of automation on general cargo transport at the logistics network level. For assessing the impacts of autonomous vehicles on logistics network structures and on freight transport routes ex-ante, an instrument for strategic transport and logistics network planning is needed. We develop an effective heuristic to find new facilities and adjust the network, while thereby considering the routing characteristics by tackling the large-scale location routing problem i.e. LRP. By the linked approach, we can estimate the exact logistics network and coordinates and measure exact transport distances, driving transport lead times and number of necessary vehicles on the infrastructure network. We carry out a case study-based qualitative evaluation of the new optimized network and investigate the logistics effects for the food retail distribution in Germany. In fact, this research reveals that the utilization of autonomous vehicles significantly enhances transportation ranges and the number of tours, while reducing the number of operating facilities.
Network Design Under Different Autonomous Truck Deployment Scenarios
Lama Al Hajj Hassan, Northwestern UniversityShow Abstract
Mike Hewitt, Loyola University
Hani Mahmassani, Northwestern University
The integration of autonomous trucks (AT) into the overall truck fleet is expected to impact industry regulations and ease driver-related challenges. It also has the potential to improve road safety and reduce carrier costs. This paper develops and tests modified service network design formulations that account for five levels of truck automation from fully manual to fully autonomous. The experiments show that cost savings with AT deployment result from partial or complete elimination of driver and sleeper team costs, reduction in empty miles traveled and the decrease in the number of trucks required to service the loads. Deployment of ATs is preferred over the long-haul direct trips connecting terminals. AT deployment even when restricted to specific geographic zones has operational benefits and environmental advantages driven by the reduction in costs and decrease in unproductive miles driven.
Capacitated Location-Allocation-Routing Problem with Time Window for On-demand Urban Air Mobility
Haleh Ale-Ahmad, Northwestern UniversityShow Abstract
Hani Mahmassani, Northwestern University
On-demand Urban Air Mobility (UAM) has gained traction with the emergence of electric propulsion with Vertical Take-Off and Landing (eVTOL). Technology companies and major aircraft manufacturers are pursuing the possibility of operating on-demand UAM at scale on a regional level and at an affordable price. However, considerable uncertainty remains about several strategic, tactical, and operational aspects that affect the viability of this business model. Assuming that the UAM operator conducts the first and last mile of the trips on the ground as well, we introduce the concept of flexible meeting points for UAM operation, where the passengers are flexible in their pick-up and drop-off locations within reasonable access/egress time. Consequently, we model UAM as a location-allocation-routing problem with time windows and present a Mixed Integer Programming formulation. The formulation addresses the request acceptance/rejection decisions, the allocation of requests to flights, demand consolidation, and routing and scheduling of the aircraft. Additionally, the formulation allows for consolidating the demand to increase the aircraft’s utilization and service rate. Depending on the operator’s business model, the proposed formulation could be used offline in a static and deterministic setting when all requests are known in advance, or it could be implemented online by dynamically solving the static and determinist snapshot problems, knowing all the requests at each decision epoch and with no knowledge about future requests.
A General Equilibrium Model for Integrated CAV Ridesourcing and Transit Services for the Morning Commute
Rong Fan, University of WashingtonShow Abstract
Daniel McCabe, University of Washington
Xuegang (Jeff) Ban (email@example.com), University of Washington
Commuting congestion increases along side the prosperity of urban cities. With the rapid development of ridesourcing services and the advances of the connected and automated vehicles (CAV), researchers are seeking innovative approaches to alleviate commuting congestion by integrating CAV-based ridesourcing and transit services. We propose a general equilibrium model for an integrated, multimodal CAV ridesourcing and transit system. Our model captures the economic behaviors and interactions of the major players (i.e.. the ridesourcing company and customers) in the commuting problem by optimizing the profit of the ridesourcing company and the utility of customers, as well as considering the network congestion. Results show that the demand for shared rides and transit are affected by the relative costs of different types of travel modes of the integrated system. While transit uses generally reduces congestion, ridesharing alone may still cause higher congestion compared with solo driving because of the deadhead miles. Our model can systematically investigate the mode choices of customers and measure the resulting congestion effect in a multimodal network, which helps bring valuable insights to transportation planers, transit agencies, and ridesourcing companies.
Autonomous Vehicle Adoption Modeling and Shared Operations Simulation: A Combined Diffusion and Traffic Microsimulation Modeling Approach
MD Jahedul Alam, Dalhousie UniversityShow Abstract
Hasan Shahrier, Dalhousie University
Muhammad Habib (firstname.lastname@example.org), Dalhousie University
This study presents a comprehensive framework to predict autonomous vehicle (AV) adoption for a traffic microsimulation model of shared operations in Halifax, Canada. A Bass Diffusion model is developed to forecast AV adoption for fulfilling the demand obtained from an activity-based model, shorter-tern Decisions Simulator (SDS). A rule-based SAV operation model is developed within a MATLAB platform to allocate SAVs to trip requests in the morning peak period. A traffic microsimulation model implements a dynamic traffic assignment process to assign SAV trips in the network and to estimate traffic impacts. The results from Bass diffusion model predict a fleet of 900 SAVs depending on the consumer’s interest to adopt. This fleet is derived based on a conservative market assumption. A second scenario of 1800 SAVs is considered, in which private companies are also assumed to be operating in Halifax. The results form SAVs assignment reveals that fleet based on a conservative market infiltration, can only serve 82% of travel demand while the second fleet, derived based on an aggressive market concept, serves 100% of the demand. The simulation results exhibit that SAVs operation in Halifax road network increases vehicle kilometer travelled (VKT) by 1% and 4% for Market Type-1(900 fleets) and Market Type-2 (1800 fleets) respectively. The empty trips are also increased by 2% and 8% for both fleets, respectively. The results of this study provide critical policy insights including the necessity of developing SAV operational strategies to minimize the number of empty trips and VKT in the network.
Surrogate-based optimization of activity-based connected subgraph problem for robotic taxi fleet service region design
Jinkai Zhou, New York UniversityShow Abstract
Joseph Chow, New York University
It is challenging to evaluate emerging transportation technologies like shared autonomous vehicle fleets on their impact on demand when designing the service region due to lack of deployment data. In optimizing the design of such regions, we propose an activity-based connected subgraph problem formulation that determines which zones to deploy a mobility service, subject to activity-based market equilibrium conditions modeled with multiagent simulation. Due to the high computational cost of such simulation, we test out the use of surrogate-based optimization applied to this complex bilevel optimization problem. Computational tests of surrogate-based method on the upper level problem with a synthetic 25-node network indicate a sufficiently low optimality gap. The algorithm is then tested on 55 zones in New York City where the lower level simulation is computed using MATSim-NYC and a 0.1%-scaled population of 8M travelers who travel by 7 different modes plus the autonomous vehicle fleet if within the service region. The algorithm is tested with budgets of 20 and 30 zones and a fleet of 50,000 vehicles (comparable to the existing taxi fleet). The optimal deployment for a 30-zone budget was 22 zones, which shows the sensitivity of the algorithm to the MATSim evaluation (more zones deployed given a fixed fleet results in longer wait times to serve customers). Compared to a benchmark 2000-iteration genetic algorithm, the surrogate algorithm can reach the same objective value within 180 iterations.
DISCLAIMER: All information shared in the TRB Annual Meeting Online Program is subject to change without notice. Changes, if necessary, will be updated in the Online Program and this page is the final authority on schedule information.