Showing papers in "Transportation Research Part C-emerging Technologies in 2021"

Risk assessment based collision avoidance decision-making for autonomous vehicles in multi-scenarios

Abstract: In this paper, we proposed a new risk assessment based decision-making algorithm to guarantee collision avoidance in multi-scenarios for autonomous vehicles. A probabilistic-model based situation assessment module using conditional random field was proposed to assess the risk level of surrounding traffic participants. Based on the assessed risk from the situation assessment module, a collision avoidance strategy with driving style preferences (e.g., aggressive or conservative) was proposed to meet the demands of different drivers or passengers. Finally, we conducted experiments in Carla (car learning to act) to evaluate our developed collision avoidance decision-making algorithm in different scenarios. The results show that our developed method was sufficiently reliable for autonomous vehicles to avoid collisions in multi-scenarios with different driving style preferences. Our developed method with adjustable driving style preferences to meet the demand of different consumers would improve drivers’ acceptance of autonomous vehicles.

Urban air mobility: A comprehensive review and comparative analysis with autonomous and electric ground transportation for informing future research

Abstract: Urban air mobility (UAM), if successful, will disrupt urban transportation. UAM is not the first disruptive technology in transportation, with recent examples including electric ground vehicles (EVs), autonomous ground vehicles (AVs), and sharing services. In this paper, we conduct a meta-analysis of about 800 articles in the UAM, EV, and AV areas that have been published from January 2015 to June 2020, and compare and contrast research thrusts in order to inform future UAM research. Alongside this effort, we conduct an in-depth review of articles related to demand modeling, operations, and integration with existing infrastructure. We use insights from the meta-analysis and comprehensive review to inform future UAM research directions. Some of the potential research directions we identify include: (1) developing more refined demand models that incorporate the timing of when individuals will adopt UAM; (2) developing high-fidelity simulation models for UAM operations that capture interactions among vertiport locations, vertiport topology, demand, pricing, dispatching, and airspace restrictions; (3) explicitly considering one-way demand and parking constraints in demand and operational models; and (4) developing more realistic time-of-day energy profiles for UAM vehicles in order to assess whether the current electrical grid can support UAM operations.

OpenACC. An open database of car-following experiments to study the properties of commercial ACC systems

Abstract: Adaptive Cruise Control (ACC) systems are becoming increasingly available as a standard equipment in modern commercial vehicles. Their penetration rate in the fleet is constantly increasing, as well as their use, especially under freeway conditions. At the same time, limited information is openly available on how these systems actually operate and their differences depending on the vehicle manufacturer or model. This represents an important gap because as the number of ACC vehicles on the road increases, traffic dynamics on freeways may change accordingly, and new collective phenomena, which are only marginally known at present, could emerge. Yet, as ACC systems are introduced as comfort options and their operation is entirely under the responsibility of the driver, vehicle manufacturers do not have explicit requirements to fulfill nor they have to provide any evidence about their performances. As a result, any safety implication connected to their interactions with other road users escapes any monitoring and opportunity of improvement. This work presents a set of experimental car-following campaigns, providing an overview of the behavior of commercial ACC systems under different driving conditions. Furthermore, the suggestion of a unified data structure across the different tests facilitates comparison between the different campaigns, vehicles, systems and specifications. The complete data is published as an open-access database (OpenACC), available to the research community. As more test campaigns will be carried out, OpenACC will evolve accordingly. The activity is performed in the framework of the openData policy of the European Commission Joint Research Centre with the objective to engage the whole scientific community towards a better understanding of the properties of ACC vehicles in view of anticipating their possible impacts on traffic flow and prevent possible problems connected to their widespread introduction. In this light, OpenACC, over time, also aims at becoming a reference point to study if and how the parameters of such systems need to be regulated, how homogeneously they behave, how new ACC car-following models should be designed for traffic microsimulation purposes and what are the key differences between ACC systems and human drivers.

A big-data driven approach to analyzing and modeling human mobility trend under non-pharmaceutical interventions during COVID-19 pandemic.

Abstract: During the unprecedented coronavirus disease 2019 (COVID-19) challenge, non-pharmaceutical interventions became a widely adopted strategy to limit physical movements and interactions to mitigate virus transmissions. For situational awareness and decision-support, quickly available yet accurate big-data analytics about human mobility and social distancing is invaluable to agencies and decision-makers. This paper presents a big-data-driven analytical framework that ingests terabytes of data on a daily basis and quantitatively assesses the human mobility trend during COVID-19. Using mobile device location data of over 150 million monthly active samples in the United States (U.S.), the study successfully measures human mobility with three main metrics at the county level: daily average number of trips per person; daily average person-miles traveled; and daily percentage of residents staying home. A set of generalized additive mixed models is employed to disentangle the policy effect on human mobility from other confounding effects including virus effect, socio-demographic effect, weather effect, industry effect, and spatiotemporal autocorrelation. Results reveal the policy plays a limited, time-decreasing, and region-specific effect on human movement. The stay-at-home orders only contribute to a 3.5%-7.9% decrease in human mobility, while the reopening guidelines lead to a 1.6%-5.2% mobility increase. Results also indicate a reasonable spatial heterogeneity among the U.S. counties, wherein the number of confirmed COVID-19 cases, income levels, industry structure, age and racial distribution play important roles. The data informatics generated by the framework are made available to the public for a timely understanding of mobility trends and policy effects, as well as for time-sensitive decision support to further contain the spread of the virus.

An analytical optimal control approach for virtually coupled high-speed trains with local and string stability

Abstract: This paper presents an analytic optimal control method for the virtually coupled train set (VCTS) in high-speed railway, aiming at maintaining consistent speed and safe spacing among trains in the VCTS. The proposed control strategy focuses on both local and string stability under variant maneuvers in the high-speed scenarios. Specifically, a state-space model is firstly formulated to describe the virtually coupled train dynamics, based on which an optimal control formulation is then constructed considering constraints of safe spacing, operation limits and train dynamic performance. To solve the proposed constrained optimal control problem, an analytical algorithm is given based on Pontryagin’s maximum principle. Further, local and string stability are analyzed, and sufficient conditions of stability are mathematically derived to guarantee stable control for both homogeneous and heterogenous VCTS. Numerical simulations were conducted to verify the correctness of derived sufficient stability conditions and the effectiveness of the proposed control strategy under variant maneuvers and disturbances.

A survey on autonomous vehicle control in the era of mixed-autonomy: From physics-based to AI-guided driving policy learning

Abstract: This paper serves as an introduction and overview of the potentially useful models and methodologies from artificial intelligence (AI) into the field of transportation engineering for autonomous vehicle (AV) control in the era of mixed autonomy when AVs drive alongside human-driven vehicles (HV). It is the first-of-its-kind survey paper to comprehensively review literature in both transportation engineering and AI for mixed traffic modeling. We will discuss state-of-the-art applications of AI-guided methods, identify opportunities and obstacles, and raise open questions. We divide the stage of AV deployment into four phases: the pure HVs, the HV-dominated, the AV-dominated, and the pure AVs. This paper is primarily focused on the latter three phases. Models used for each phase are summarized, encompassing game theory, deep (reinforcement) learning, and imitation learning. While reviewing the methodologies, we primarily focus on the following research questions: (1) What scalable driving policies are to control a large number of AVs in mixed traffic comprised of human drivers and uncontrollable AVs? (2) How do we estimate human driver behaviors? (3) How should the driving behavior of uncontrollable AVs be modeled in the environment? (4) How are the interactions between human drivers and autonomous vehicles characterized? We also provide a list of public datasets and simulation software related to AVs. Hopefully this paper will not only inspire our transportation community to rethink the conventional models that are developed in the data-shortage era, but also start conversations with other disciplines, in particular robotics and machine learning, to join forces towards creating a safe and efficient mixed traffic ecosystem.

Explaining shared micromobility usage, competition and mode choice by modelling empirical data from Zurich, Switzerland

Abstract: Shared micromobility services (e-scooters, bikes, e-bikes) have rapidly gained popularity in the past few years, yet little is known about their usage. While most previous studies have analysed single modes, only few comparative studies of two modes exist and none so-far have analysed competition or mode choice at a high spatiotemporal resolution for more than two modes. To this end, we develop a generally applicable methodology to model and analyse shared micromobility competition and mode choice using widely accessible vehicle location data. We apply this methodology to estimate the first comprehensive mode choice models between four different micromobility modes using the largest and densest empirical shared micromobility dataset to-date. Our results suggest that mode choice is nested (dockless and docked) and dominated by distance and time of day. Docked modes are preferred for commuting. Hence, docking infrastructure for currently dockless modes could be vital for bolstering micromobility as an attractive alternative to private cars to tackle urban congestion during rush hours. Furthermore, our results reveal a fundamental relationship between fleet density and usage. A “plateau effect” is observed with decreasing marginal utility gains for increasing fleet densities. City authorities and service providers can leverage this quantitative relationship to develop evidence-based micromobility regulation and optimise their fleet deployment, respectively.

Predicting origin-destination ride-sourcing demand with a spatio-temporal encoder-decoder residual multi-graph convolutional network

Abstract: With the rapid development of mobile-internet technologies, on-demand ride-sourcing services have become increasingly popular and largely reshaped the way people travel. Demand prediction is one of the most fundamental components in supply-demand management systems of ride-sourcing platforms. With an accurate short-term prediction for origin-destination (OD) demand, the platforms make precise and timely decisions on real-time matching, idle vehicle reallocations, and ride-sharing vehicle routing, etc. Compared to the zone-based demand prediction that has been examined in many previous studies, OD-based demand prediction is more challenging. This is mainly due to the complicated spatial and temporal dependencies among the demand of different OD pairs. To overcome this challenge, we propose the Spatio-Temporal Encoder-Decoder Residual Multi-Graph Convolutional network (ST-ED-RMGC), a novel deep learning model for predicting ride-sourcing demand of various OD pairs. Firstly, the model constructs OD graphs, which utilize adjacent matrices to characterize the non-Euclidean pair-wise geographical and semantic correlations among different OD pairs. Secondly, based on the constructed graphs, a residual multi-graph convolutional (RMGC) network is designed to encode the contextual-aware spatial dependencies, and a long-short term memory (LSTM) network is used to encode the temporal dependencies, into a dense vector space. Finally, we reuse the RMGC networks to decode the compressed vector back to OD graphs and predict the future OD demand. Through extensive experiments on the for-hire-vehicles datasets in Manhattan, New York City, we show that our proposed deep learning framework outperforms the state-of-arts by a significant margin.

Mixed platoon control of automated and human-driven vehicles at a signalized intersection: Dynamical analysis and optimal control

Abstract: The emergence of Connected and Automated Vehicles (CAVs) promises better traffic mobility for future transportation systems. Existing research mostly focused on fully-autonomous scenarios, while the potential of CAV control at a mixed traffic intersection where human-driven vehicles (HDVs) also exist has been less explored. This paper proposes a notion of “ 1 + n ” mixed platoon, consisting of one leading CAV and n following HDVs, and formulates a platoon-based optimal control framework for CAV control at a signalized intersection. Based on the linearized dynamics model of the “ 1 + n ” mixed platoon, fundamental properties including stability and controllability are under rigorous theoretical analysis. Then, a constrained optimal control framework is established, aiming at improving the global traffic efficiency and fuel consumption at the intersection via direct control of the CAV. A hierarchical event-triggered algorithm is also designed for practical implementation of the optimal control method between adjacent mixed platoons when approaching the intersection. Extensive numerical simulations at multiple traffic volumes and market penetration rates validate the greater benefits of the mixed platoon based method, compared with traditional trajectory optimization methods for one single CAV.

Modeling epidemic spreading through public transit using time-varying encounter network.

Abstract: Passenger contact in public transit (PT) networks can be a key mediate in the spreading of infectious diseases. This paper proposes a time-varying weighted PT encounter network to model the spreading of infectious diseases through the PT systems. Social activity contacts at both local and global levels are also considered. We select the epidemiological characteristics of coronavirus disease 2019 (COVID-19) as a case study along with smart card data from Singapore to illustrate the model at the metropolitan level. A scalable and lightweight theoretical framework is derived to capture the time-varying and heterogeneous network structures, which enables to solve the problem at the whole population level with low computational costs. Different control policies from both the public health side and the transportation side are evaluated. We find that people's preventative behavior is one of the most effective measures to control the spreading of epidemics. From the transportation side, partial closure of bus routes helps to slow down but cannot fully contain the spreading of epidemics. Identifying "influential passengers" using the smart card data and isolating them at an early stage can also effectively reduce the epidemic spreading.

Automated vehicle-involved traffic flow studies: A survey of assumptions, models, speculations, and perspectives

Abstract: Automated vehicles (AVs) are widely considered to play a crucial role in future transportation systems because of their speculated capabilities in improving road safety, saving energy consumption, reducing vehicle emissions, increasing road capacity, and stabilizing traffic. To materialize these widely expected potentials of AVs, a sound understanding of AVs’ impacts on traffic flow is essential. Not surprisingly, despite the relatively short history of AVs, there have been numerous studies in the literature focusing on understanding and modeling various aspects of AV-involved traffic flow and significant progresses have already been made. To understand the recent development and ultimately inspire new research ideas on this important topic, this survey systematically and comprehensively reviews the existing AV-involved traffic flow models with different levels of details, and examines the relationship among the design of AV-based driving strategies, the management of transportation systems, and the resulting traffic dynamics. The pros and cons of the existing models and approaches are critically discussed, and future research directions are also provided.

Macroscopic modeling and dynamic control of on-street cruising-for-parking of autonomous vehicles in a multi-region urban road network

Abstract: The spatio-temporal imbalance of parking demand and supply results in unwanted on-street cruising-for-parking traffic of conventional vehicles. Autonomous vehicles (AVs) can self-relocate to alleviate the shortage of parking supplies at the trip destinations. The extra floating trips of vacant AVs have adverse impacts on traffic congestion and the parking demand–supply imbalance may still exist when they are not distributed optimally. This paper presents a centralized parking dispatch approach to optimize the distribution of floating AVs and provide regional route guidance. We apply the concept of macroscopic fundamental diagram to represent the evolution of traffic conditions, cruising-for-parking, and dispatched AVs in a congested multi-region network. A model predictive control is suggested to optimize the control inputs. Numerical experiments in a four-region network demonstrate that the proposed parking dispatch and regional route guidance of AVs are effective in reducing intense cruising-for-parking traffic, and the integration of both has the best control performance by regulating the network towards under-saturated conditions. The performance of the proposed schemes is evaluated via simulations with noise in measurement errors and compliance rate prediction. Results show substantial improvements in terms of total time spent, even for low levels of AV market penetration or AV compliance rate to parking dispatch and route guidance.

Multi-community passenger demand prediction at region level based on spatio-temporal graph convolutional network

Abstract: Region-level passenger demand prediction plays an important role in the coordination of travel demand and supply in the urban public transportation system. The complex urban road network structure leads to irregular shapes and arrangements of regions, which poses a challenge for capturing the spatio-temporal correlation of demand generated in different regions. In this study, we propose a multi-community spatio-temporal graph convolutional network (MC_STGCN) framework to predict passenger demand at a multi-region level by exploring spatio-temporal correlations among regions. Specifically, the gated recurrent unit (GRU) is applied to encode the temporal correlation in regions into a vector. On the other hand, the spatial correlations among regions are encoded into two graphs through the graph convolutional network (GCN): geographically adjacent graph and functional similarity graph. Then, a prediction module based on the Louvain algorithm is used to accomplish the passenger demand prediction of multi-regions. The two real-world taxi order data collected in Shenzhen City and New York City are used in model validation and comparison. The numerical results show that the MC_STGCN model outperforms both classical time-series prediction methods and deep learning approaches. Moreover, in order to better illustrate the superiority of the proposed model, we further discuss the improvement of prediction performance though spatio-temporal correlation modeling and analyzing, the effectiveness of community detection compared with random classification of regions, and the advantages of regional level prediction compared with grid-based prediction models.

Large-scale pavement roughness measurements with vehicle crowdsourced data using semi-supervised learning

Abstract: Rapid measurements of large-scale pavement roughness have long been a hot topic in pavement condition evaluation and maintenance. Most traditional methods rely on dedicated devices, such as laser, Lidar and so on, which should be set up on customized vehicles. With the rapid development of sensing technology, vehicles owned by the general public are empowered with the ability to collect vibration measurements themselves. This crowdsourced dataset is convenient, extensive coverage, inexpensive, and has high sampling frequency, making it a suitable source for large-scale pavement roughness evaluation. However, vehicle information is missing for these data due to privacy protection, which renders them quite difficult to directly use with traditional model-based methods. Thus, in this paper, we propose a semi-supervised learning (SSL) model to deal with the problem of incomplete data and multi-vehicle data fusion. A mathematical derivation of the ‘international roughness index’ (IRI) using in-car vibrations is established. Furthermore, given the multi-vehicle scenario, a self-training model is designed to iteratively estimate IRIs in a roadway network. Both the confidences of the vehicle parameters and IRI estimation are considered in the algorithm to improve its reliability and robustness. A full-car simulation model is constructed to verify the effectiveness of the proposed model. The results show that the overall relative error is less than 10% for 50 road sections in the network, which is a significant improvement compared to traditional multi-vehicle average models. The errors of the SSL model are found to be significantly dependent on the iteration order. Based on the proposed model, the coupled impact of the sampling rate and vehicle quantity on the model’s accuracy is further discussed. The proposed approach provides new insights into large-scale pavement roughness measurements.

On the inefficiency of ride-sourcing services towards urban congestion

Abstract: The advent of shared-economy and smartphones made on-demand transportation services possible, which created additional opportunities, but also more complexity to urban mobility. Companies that offer these services are called Transportation Network Companies (TNCs) due to their internet-based nature. Although ride-sourcing is the most notorious service TNCs provide, little is known about to what degree its operations can interfere in traffic conditions, while replacing other transportation modes, or when a large number of idle vehicles is cruising for passengers. We experimentally analyze the efficiency of TNCs using taxi trip data from a Chinese megacity and an agent-based simulation with a trip-based MFD model for determining the speed. We investigate the effect of expanding fleet sizes for TNCs, passengers’ inclination towards sharing rides, and strategies to alleviate urban congestion. We observe that, although a larger fleet size reduces waiting time, it also intensifies congestion, which, in turn, prolongs the total travel time. Such congestion effect is so significant that it is nearly insensitive to passengers’ willingness to share and flexible supply. Finally, parking management strategies can prevent idle vehicles from cruising without assigned passengers, mitigating the negative impacts of ride-sourcing over congestion, and improving the service quality.

Classifying travelers' driving style using basic safety messages generated by connected vehicles: application of unsupervised machine learning

Abstract: Driving style can substantially impact mobility, safety, energy consumption, and vehicle emissions. While a range of methods has been used in the past for driving style classification, the emergence of connected vehicles equipped with communication devices provides a new opportunity to classify driving style using high-resolution (10 Hz) microscopic real-world data. In this study, location-based big data and machine learning are used to classify driving styles ranging from aggressive to calm. This classification can be used to customize driver assistance systems, assess mobility, crash risk, fuel consumption, and emissions. This study’s main objective is to develop a framework that harnesses Basic Safety Messages (BSMs) generated by connected vehicles to quantify instantaneous driving behavior and classify driving styles in different spatial contexts using unsupervised machine learning methods. To this end, a subset of the Safety Pilot Model Deployment (SPMD) with more than 27 million BSM observations generated by more than 1300 individuals making trips on diverse roadways and through several neighborhoods in Ann Arbor, Michigan, were processed and analyzed. To quantify driving style, the concept of temporal driving volatility, as a surrogate safety measure of unsafe driving behavior, was utilized and applied to vehicle kinematics, i.e., observed speeds and longitudinal/lateral accelerations. Specifically, six volatility measures are extracted and used for classifying drivers. K-means and K-medoids methods are applied for grouping drivers in aggressive, normal, and calm clusters. Clustering results indicate that not only does driving style vary among drivers, but the thresholds for aggressive and calm driving vary across different roadway types due to variations in environment and road conditions. The proportion of aggressive driving styles was also higher on commercial streets than on highways and residential streets. Notably, we propose a Driving Score to measure driving performance consistently across drivers.

A DQN-based intelligent control method for heavy haul trains on long steep downhill section

Abstract: The cyclic air braking strategy on the long and steep downhill section is one of the biggest challenges for heavy haul railway lines in China. To deal with this problem, this paper presents an intelligent control method based on the Deep-Q-Network (DQN) algorithm for the heavy haul train running on the long and steep downhill section. The aim of the optimal train control problem in the paper is to enhance the train operation performance in regard to the operational safety, maintenance costs and running efficiency. In the train control model, the characteristics of the heavy haul train, the speed limits and constraints on the air-refilling time of the train pipe are taken into consideration. Then the train control process on the long and steep downhill section is described as a Markov decision process for the application of the reinforcement learning (RL) technique. Further, the critical elements of RL are designed and an intelligent control method on the basis of the DQN algorithm is developed to address the optimal train control problem in this paper. Finally, experimental simulations are carried out with the actual data of the Shuozhou-Huanghua Line such that the effectiveness and robustness of the proposed DQN-based control method are verified.

Hybrid deep reinforcement learning based eco-driving for low-level connected and automated vehicles along signalized corridors

Abstract: Eco-Driving has great potential in reducing the fuel consumption of road vehicles, especially under the connected and automated vehicles (CAVs) environment. Traditional model-based Eco-Driving methods usually require sophisticated models and cannot deal with complex driving scenarios. This paper proposes a hybrid reinforcement learning (RL) based Eco-Driving algorithm considering both the longitudinal acceleration/deceleration and the lateral lane-changing operations. A deep deterministic policy gradient (DDPG) algorithm is designed to learn the continuous longitudinal acceleration/deceleration to reduce the fuel consumption as well as to maintain acceptable travel time. Collecting the critic’s value of each single lane from DDPG and integrating the information of adjacent lanes, a deep Q-learning algorithm is developed to make the discrete lane-changing decision. Together, a hybrid deep Q-learning and policy gradient (HDQPG) method is developed for vehicles driving along multi-lane urban signalized corridors. The method can enable the controlled vehicle to learn well-established longitudinal fuel-saving strategies, and to perform appropriate lane-changing operations at proper times to avoid congested lanes. Numerical experiments show that HDQPG can reduce fuel consumption by up to 46% with marginal or no increase of travel times.

Short-term origin-destination demand prediction in urban rail transit systems: A channel-wise attentive split-convolutional neural network method

Abstract: Short-term origin–destination (OD) flow prediction in urban rail transit (URT) plays a crucial role in smart and real-time URT operation and management. Different from other short-term traffic forecasting methods, the short-term OD flow prediction possesses three unique characteristics: (1) data availability: real-time OD flow is not available during the prediction; (2) data dimensionality: the dimension of the OD flow is much higher than the cardinality of transportation networks; (3) data sparsity: URT OD flow is spatiotemporally sparse. There is a great need to develop novel OD flow forecasting method that explicitly considers the unique characteristics of the URT system. To this end, a channel-wise attentive split–convolutional neural network (CAS-CNN) is proposed. The proposed model consists of many novel components such as the channel-wise attention mechanism and split CNN. In particular, an inflow/outflow-gated mechanism is innovatively introduced to address the data availability issue. We further originally propose a masked loss function to solve the data dimensionality and data sparsity issues. The model interpretability is also discussed in detail. The CAS–CNN model is tested on two large-scale real-world datasets from Beijing Subway, and it outperforms the rest of benchmarking methods. The proposed model contributes to the development of short-term OD flow prediction, and it also lays the foundations of real-time URT operation and management.

On-demand ridesharing with optimized pick-up and drop-off walking locations

Abstract: On-demand systems in which passengers with similar routes can share a vehicle are expected to become a relevant part of future mobility, thanks to their flexibility and their potential impact on reducing congestion. Nevertheless, due to the long detours required by a door-to-door scheme, they induce extra costs to the users in terms of delay. In this paper, we face the design of such a system in which users might be requested online to walk towards/from nearby pick-up/drop-off points if this improves overall efficiency. We show theoretically that the general problem becomes more complex (as it contains two sub-problems that extend set-cover), analyze the trade-offs that emerge, and provide a general formulation and specific heuristics that are able to solve it over large instances. We test this formulation over a real dataset of Manhattan taxi trips (9970 requests during one hour), finding that (a) average walks of about one minute can reduce the number of rejections in more than 80% and Vehicles-Hour-Traveled in more than 10%, (b) users who depart or arrive at the most demanded areas are more likely to be required to walk, and (c) the performance improvement of the service is larger when the system receives more trip requests.

TrajGAIL: Generating urban vehicle trajectories using generative adversarial imitation learning

Abstract: Recently, an abundant amount of urban vehicle trajectory data has been collected in road networks. Many studies have used machine learning algorithms to analyze patterns in vehicle trajectories to predict location sequences of individual travelers. Unlike the previous studies that used a discriminative modeling approach, this research suggests a generative modeling approach to learn the underlying distributions of urban vehicle trajectory data. A generative model for urban vehicle trajectories can better generalize from training data by learning the underlying distribution of the training data and, thus, produce synthetic vehicle trajectories similar to real vehicle trajectories with limited observations. Synthetic trajectories can provide solutions to data sparsity or data privacy issues in using location data. This research proposes TrajGAIL, a generative adversarial imitation learning framework for the urban vehicle trajectory generation. In TrajGAIL, learning location sequences in observed trajectories is formulated as an imitation learning problem in a partially observable Markov decision process. The model is trained by the generative adversarial framework, which uses the reward function from the adversarial discriminator. The model is tested with both simulation and real-world datasets, and the results show that the proposed model obtained significant performance gains compared to existing models in sequence modeling.

Calibration and evaluation of responsibility-sensitive safety (RSS) in automated vehicle performance during cut-in scenarios

Abstract: The ability of automated vehicles (AV) to avoid accidents in complex traffic environments is the focus of considerable public attention. Intel has proposed a mathematical model called Responsibility-Sensitive Safety (RSS) to ensure AVs maintain a safe distance from surrounding vehicles, but testing has, to date, been limited. This study calibrates and evaluates the RSS model based on cut-in scenarios in which minimal time-to-collision (TTC) is less than 3 s. Two hundred cut-in events were extracted from Shanghai Naturalistic Driving Study data, and the corresponding scenario information for each event was imported into a simulation platform. In each scenario, the human driver was replaced by an AV driven by the model predictive control-based adaptive cruise control (ACC) system embedded with the RSS model. The safety performance of three conditions, the human driver, RSS-embedded ACC model, and ACC-only model, were evaluated and compared. Compared to the performance of human drivers and ACC-only algorithm respectively, the RSS model increased the average TTC per event by 2.86 s and 0.94 s, shortened time-exposed TTC by 1.34 s and 0.65 s, and reduced time-integrated TTC by 0.91 s2 and 0.72 s2. These changes indicate that the RSS-embedded ACC model can improve safety performance in emergent cut-in scenarios. The RSS model can therefore be applied as a security guarantee, that is, to ensure the AV’s timely awareness and response to dangerous cut-in situations, thus mitigating potential conflict.

Requiem on the positive effects of commercial adaptive cruise control on motorway traffic and recommendations for future automated driving systems

Abstract: Connected and automated vehicles (CAVs) promise to significantly improve road traffic. To a certain extent, this situation is similar to the expectations at the end of the last century about the positive effects that the introduction of Adaptive Cruise Control (ACC) systems would have had on motorway traffic. The parallelism is interesting because ACC equipped vehicles represent the first level of vehicle automation and are now widely available on the market. In this light, studying ACC impacts can help to anticipate potential problems related to its widespread application and to avoid that AVs and CAVs will lead to the same problems. Only a few test-campaigns had been carried out studying the ACC impacts under real-world driving conditions in quantitative terms. To bridge this gap, the Joint Research Centre of the European Commission has organized a number of experimental campaigns involving several ACC-equipped vehicles to study different implications of their widespread. In this context, the present paper summarizes the outcomes of a test campaign involving 10 commercially available ACC-equipped vehicles. The test campaign has been executed in two different test-tracks of the ZalaZONE proving ground, in Hungary. The tests have been carried out at low-speeds (30–60 km/h) and have involved platoons of vehicles of different brands and different powertrains, which were tested in a variety of vehicle orders and with different settings of their ACC systems. Test results have been used to derive information about the properties of the different ACC systems, to study their string stability, to study the effect of ACC systems on traffic flow, and to draw inference about the possible implications on energy consumption and traffic safety. Results confirm the previous findings in terms of string instability of the ACC and highlight that in the present form, ACC systems will possibly lead to higher energy consumption and introduce new safety risks when their penetration in the fleet increases. However, they also highlight that the materialization of the above findings for AVs depends on the operational logic that manufacturers will adopt during the implementation phase. Therefore, results suggest that functional requirements to guarantee string stability and in general to not disrupt the normal flow of traffic should be introduced both for ACC and for any automated system that will be placed on the market in the future.

The Multi-visit Traveling Salesman Problem with Multi-Drones

Abstract: The use of drones for parcel delivery has recently attracted wide attention due to its potential in improving efficiency of the last-mile delivery. Though attempts have been made on combined truck-drone delivery to deploy multiple drones that can deliver multiple packages per trip, many placed extra assumptions to simplify the problem. This paper investigates the multi-visit traveling salesman problem with multi-drones (MTSP-MD), whose objective is to minimize the time (makespan) required by the truck and the drones to serve all customers together. The energy consumption of the drone depends on the flight time, the self-weight of the drone and the total weight of packages carried by the drone, which declines after each delivery throughout the drone flight. The MTSP-MD problem consists of three complicated sub-problems, namely (1) the drone flight problem with both a payload capacity constraint and an energy endurance constraint, (2) the traveling salesman problem with precedence constraints, and (3) the synchronization problem between the truck route and the drone schedules. The problem is first formulated into a mixed-integer linear program (MILP) model and we propose a multi-start tabu search (MSTS) algorithm with tailored neighborhood structure and a two-level solution evaluation method that incorporates a drone-level segment-based evaluation and a solution-level evaluation based on the critical path method (CPM). The experimental results demonstrate the accuracy and efficiency of our proposed algorithm on small-scale instances and show a significant cost reduction when considering multi-visits, multi-drones, and drones with higher payload capacity and higher battery capacity for medium and large-scale instances.

Adaptive Traffic Signal Control for large-scale scenario with Cooperative Group-based Multi-agent reinforcement learning

Abstract: Recent research reveals that reinforcement learning can potentially perform optimal decision-making compared to traditional methods like Adaptive Traffic Signal Control (ATSC). With the development of knowledge through trial and error, the Deep Reinforcement Learning (DRL) technique shows its feasibility for the intelligent traffic lights control. However, the general DRL algorithms cannot meet the demands of agents for coordination within large complex road networks. In this article, we introduce a new Cooperative Group-Based Multi-Agent reinforcement learning-ATSC (CGB-MATSC) framework. It is based on Cooperative Vehicle Infrastructure System (CVIS) to realize effective control in the large-scale road network. We propose a CGB-MAQL algorithm that applies k -nearest-neighbor-based state representation, pheromone-based regional green-wave control mode, and spatial discounted reward to stabilize the learning convergence. Extensive experiments and ablation studies of the CGB-MAQL algorithm show its effectiveness and scalability in the synthetic road network, Monaco city and Harbin city scenarios. Results demonstrate that compared with a set of general control methods, our algorithm can better control multiple intersection cases on congestion alleviation and environmental protection.

About calibration of car-following dynamics of automated and human-driven vehicles: Methodology, guidelines and codes

Abstract: A comprehensive literature review reveals that there exist lots of ambiguities, confusion and even contradictions in setting a car-following calibration problem. In particular, confusion arises in the selection of measure of performances and goodness-of-fit functions. In this study, a methodology to compare and rank objective functions is thus proposed, which is based on Pareto-efficiency and on indifference curves. The methodology has been applied to all objective functions used in the field literature so far (and to new ones), in a vast set of calibration experiments. The experiments involved two car-following models and two adaptive cruise control (ACC) algorithms, and four different datasets, including both automated and human-driven vehicles trajectories. Since results are consistent among all the calibration experiments, a sound and robust guideline to calibrate car-following dynamics has been proposed. It includes recommendation about what calibration settings should be avoided and what are to be adopted. On the one hand, a general agreement on a sound calibration setting for car-following models is deemed necessary for comparing results from different studies which use different models and datasets. On the other hand, any new car-following model or objective function being developed in the future shall be compared with existing ones in a fair and impartial manner. For these reasons, and to promote and enable transparent and reproducible research, codes and data from this study are shared with the community.

Domain Adaptation from Daytime to Nighttime: A Situation-sensitive Vehicle Detection and Traffic Flow Parameter Estimation Framework

Abstract: Vehicle detection in traffic surveillance images is an important approach to obtain vehicle data and rich traffic flow parameters. Recently, deep learning based methods have been widely used in vehicle detection with high accuracy and efficiency. However, deep learning based methods require a large number of manually labeled ground truths (bounding box of each vehicle in each image) to train the Convolutional Neural Networks (CNN). In the modern urban surveillance cameras, there are already many manually labeled ground truths in daytime images for training CNN, while there are little or much less manually labeled ground truths in nighttime images. In this paper, we focus on the research to make maximum usage of labeled daytime images (Source Domain) to help the vehicle detection in unlabeled nighttime images (Target Domain). For this purpose, we propose a new situation-sensitive method based on Faster R-CNN with Domain Adaptation (DA) to improve the vehicle detection at nighttime. Furthermore, a situation-sensitive traffic flow parameter estimation method is developed based on the traffic flow theory. We collected a new dataset of 2,200 traffic images (1,200 for daytime and 1,000 for nighttime) of 57,059 vehicles to evaluate the proposed method for the vehicle detection. Another new dataset with three 1,800-frame daytime videos and one 1,800-frame nighttime video of about 260 K vehicles was collected to evaluate and show the estimated traffic flow parameters in different situations. The experimental results show the accuracy and effectiveness of the proposed method.

Analytical analysis of the effect of maximum platoon size of connected and automated vehicles

Abstract: The maximum platoon size is a critical parameter in connected and automated vehicle (CAV) platoon configuration. However, the effect of platoon size on the transportation system has not been well-studied. This paper unveils the effect of maximum CAV platoon size in terms of road capacity and traffic flow stability. Specifically, the analytical formulations of the capacity and flow stability are developed considering the maximum platoon size. Simulations are conducted to verify the developed theoretical models. For capacity analysis, both the analytical and simulation results indicate that a larger maximum platoon size can help increase the capacity. However, the increment becomes smaller with the increase of maximum platoon size. For flow stability analysis, the theoretical analysis and microscopic simulation show that smaller maximum platoon size leads to greater traffic flow stabilization. In addition, analysis shows that improvements in capacity and traffic stability are more profound when CAV penetration and platooning intensity are high.

Simulation of price, customer behaviour and system impact for a cost-covering automated taxi system in Zurich

Abstract: Automated Mobility on Demand (AMoD) is a concept that has recently generated much discussion In cases where large-scale adoption of an automated taxi service is anticipated, the service’s impacts may become relevant to key transport system metrics, and thus to transport planners and policy-makers as well In light of this increasingly important question, this paper presents an agent-based transport simulation with (single passenger) AMoD In contrast to earlier studies, all scenario data (including demand patterns, cost assumptions and customer behaviour) is obtained for one specific area, the city of Zurich, Switzerland The simulation study fuses information from a detailed bottom-up cost analysis of mobility services in Switzerland, a specifically tailored Stated-Preferences survey about automated mobility services conducted in the canton of Zurich, and a detailed agent-based transport simulation for the city, based on MATSim Methodologically, a comprehensive approach is presented that iteratively runs these components to derive states in which service cost, waiting times and demand are in equilibrium for a cost-covering AMoD operator with predefined fleet size For Zurich, several cases are examined, with 4,000 AMoD vehicles leading to the maximum demand of around 150,000 requests per day that can be attracted by the system Within these parameters, the simulation results show that customers are willing to accept average waiting times of around 4 min at a price of 075 CHF/km Further cost-covering cases with lower demand are presented, where either smaller fleet sizes lead to higher waiting times, or larger fleet sizes lead to higher costs While our simulations indicate that an AMoD system in Zurich can bring benefits to the users, they show that the system impact is largely negative Caused by modal shifts, our simulations show an increase of driven distance of up to 100% All examined fleet configurations of the unregulated, cost-covering, single-passenger, door-to-door AMoD service are found to be highly counter-productive on a path towards a more shared and active transport system Accordingly, policy recommendations for regulation are discussed

What do we (Not) know about our future with automated vehicles

Abstract: While research on developing and testing automated vehicle (AVs) technologies is well underway, research on their implications on travel-related behavior is in its infancy The aim of this paper is to summarize and analyze literature that focuses on travel-related behavior impacts of AVs, namely levels 4 and 5, as well as highlight important directions of research We review five methods used to quantitatively investigate these implications and how each method contributes to this literature: 1) controlled testbeds, 2) driving simulators and virtual reality, 3) agent-based and travel-demand models, 4) surveys, and 5) field experiments We also present five critical research questions regarding the implications of AVs on the demand side of transportation and summarize findings from the current literature on: 1) what is the willingness to adopt the technology? and what are the impacts of the technology on 2) in-vehicle behavior? 3) value of time? 4) travel-related behaviors (activity pattern, mode, destination, residential location)? and 5) vehicle miles traveled (VMT)? Results can be divided into four categories The first category corresponds to results on research questions with numerous data points where the direction of the impact is consistent across the literature, albeit the magnitude varies considerably For instance, surveys indicate 19% to 68% of people are unwilling to adopt AV technology, a sentiment that is fading over time Moreover, people prefer owning AVs over sharing them and don’t believe their car ownership will decrease Regarding VMT, most studies predict an increase that varies from a low of 1% to a high of 90% depending on the scenario and assumptions under study The second category of findings corresponds to research questions with limited and consistent, albeit highly variable data points For example, a few stated preference survey studies indicate that reduced stress and multitasking during travel will reduce the value of time between 5% and 90% The third category of results is on research questions with a few but conflicting data points For instance, surveys indicate that people (80% to 85%) do not believe their residential location will be impacted by the adoption of AVs Some simulation studies, however, indicate that lower travel costs will drive people away from cities and into suburbs while other studies report the opposite The final category of results corresponds to research question with a single or no data points For instance, one study explores how users will use vehicles to run errands while no studies investigate user preferences for vehicle types (eg mobile-homes vs right-sized) or how they plan to use their vehicles when they are not needed (eg rent out vs leave them idle) Moving forward, the goal is to shift all results into the first category while simultaneously tightening the prediction interval of the magnitude of the impacts This can be achieved by: 1) focusing more efforts on research questions that fall under the three remaining categories to fill the holes in the literature, and 2) establishing consistency and clarity of assumptions used by researchers to enable comparisons and transferability of results