Simulation is a very useful tool that has been applied in many fields, including parking management (Zorn et al. (2011), Gallo et al. (2011) and Arnott and Inci (2006), Martens et al. (2007)). These models were macroscopic or mesoscopic, thus lacking some of the finer modeling details that might be useful in capturing specific parameters of detailed parking modeling. While Van der Waerden et al. (2002) and Caicedo et al. (2006) used microscopic models in simulating parking, most of this work is dedicated to the “normal” operations. Modeling and management of special events has seen some attention in the literature (Shao et al. (2008), Sarasua et al. (2005)). Although existing models of driver behavior do not take into account adverse disaster scenarios, they do take into account factors such as aggression (e.g. Toledo et al., 2009) and risk aversion that are known to be impacted by emotion. Behavioral aspects have been considered in traffic management applications (Antoniou et al., 2011), often in the context of adverse/emergency conditions (e.g. Balakrishna, Antoniou et al., 2008; Prionisti and Antoniou, 2012). The use of multiple, diverse technologies for localization in the context of indoor and harsh environments has seen a lot of interest in the literature recently (Addesso et al., 2010; Dedes et al., 2011; Prieto et al., 2012; Tanigawa et al., 2004) and is considered a critical source of accurate and reliable data for the applications in this research.
The objective of the EMPARCO (Efficient Management of Parking under Constraints) project is to develop solutions for the management of large-scale parking facilities and depots (for either passenger vehicles or commercial fleets) under constraints including (i) near-capacity demand, (ii) temporally concentrated arrivals/departures and (iii) need for emergency evacuation. The specific modeling requirements that need to be covered in terms of both the demand/behavioral and supply side mainly include:
- Modeling of parking spot choice and the process of finding a spot in a facility that is almost at capacity
- Modeling of parking facilities, including parking spots, access corridors and ramps (for multi-story facilities).
- Modeling of the traffic dynamics of vehicles operating in parking facilities and commercial vehicle depots, both under regular conditions, but also under stress or unusual conditions.
Regarding the localization support, existing localization techniques and methodologies for hybrid and indoor parking areas should be evaluated. This includes evaluation of GNSS- and GNSS/MEMS-based systems and radio access technologies. Finally, given a reliable simulation model of the operating facility, based on robust localization data from available sources, a response strategy should be determined, comprising (in general) (i) information to be provided to the drivers and (ii) control strategies that may be used to guide and restrict the movement of the same drivers.
An integrated methodological framework is developed that operationalizes a cycle created from the following main methodological and technological challenges/ components:
- Microscopic parking facility simulation to provide the modeling and simulation capabilities to accurately model the required scenarios for the execution of this project, both in terms of the supply- and the demand-side.
- Novel methodologies and algorithms for vehicle localization support, including data fusion of various data sources
- Information generation and dissemination and control
- Strategy generation for the optimal parking management schemes
In urban situations, with obstructions, such as tunnels, urban canyons, and indoor parking structures, GPS information may be unavailable. In this research, we have provided fields experiments, in which data were collected using low-cost, ubiquitous equipment. The collected data have been used to investigate the feasibility of using acceleration distributions for calibration. Furthermore, we have motivated the use of distributions for the calibration and validation of traffic simulation models and presented a framework that supports it. Distribution-based calibration allows us to exploit the richness of data and relax the unrealistic assumptions that are requested using the traditional way of calibration. This research has also suggested a simple methodology for identifying and classifying different driving behaviors. Nowadays, rich data obtained from opportunistic smartphone sensors allow monitoring indoor driving behavior and in combination with clustering methods could be used for the identification of certain driving behavior patterns under certain traffic conditions. Moreover, these data are useful for recognizing special situations, i.e. for anomaly detection. Multi-level modeling of an existing large-scale parking facility has been achieved in both AIMSUN and TransModeler simulation software. Some scenarios for evacuation have been demonstrated and conclusions for the contribution of guidance and control techniques have been drawn. The evacuation time could be significantly reduced (48%) by providing the appropriate guidance to the drivers either through applications in smartphones or variable message signs. The suggested plans and strategies will insure a reduction in the total evacuation time.
Traffic simulation is a mature field, with several decades of development. While some aspects can be assumed to be at a level where most challenges have been overcome, there are still aspects that remain unsolved. For example, traffic simulation of mixed networks at conditions close to or exceeding capacity are still challenging. Similarly, modeling low-speed traffic is also a challenging task (often leading to underestimation of the capacity), while parking maneuvers and their impact on the following/opposing vehicles are aspects, in which modeling can be improved. With this project, all these innovative aspects are taken into consideration. Focus is placed on the more challenging aspects of modeling vehicle operation at a microscopic scale in parking facilities, which combine a number of restrictions along the state-of-the-art of traffic modeling and simulation, i.e. complex geometry, congested conditions, and very low speeds. Gap-acceptance and merging models that are formulated/estimated for general traffic would perform poorly when applied to modeling traffic facilities. In addition, behavioral aspects and the impact of stressful driving conditions are also taken into account in this research. Last but not least, the use of multiple, diverse technologies for localization in the context of indoor and harsh environments is considered as a critical source of accurate and reliable data for the applications in this research.
In outdoor and hybrid environments emphasis is placed in MEMS INS and terrain-aiding augmentation techniques, whereas, in indoor areas priority shall be given in GNSS/MEMS INS systems aided by radio-based technologies. Rigorous evaluation of existing localization techniques in indoor parking areas for various radio access technologies (Wi-Fi, Bluetooth, 3G Femto Cells and UWB) taking into account realistic propagation effects on the signal strength. In particular, the impact of multipath, Doppler Shifts and through obstacles and materials propagation have been considered for the limitation of the localization accuracy. These effects have been accurately quantified and novel mitigation techniques are developed considering single and multi-antenna systems. Novel localization algorithms tailored for indoor parking areas (deep fades, slow variations of the signal and low speed of the wireless nodes) have been developed taking into account new approaches and capabilities of the wireless networks including cooperation techniques. These algorithms also include augmentation of GNSS/MEMS, gyroscopes and magnetometers with UWB for seamless positioning. This fusion approach will enable better control of UWB calibration errors (due to receiver clock biases), which is important, especially in microscopic simulations because of the fine level of detail required in these models. Until recently, obtaining detailed traffic information on individual vehicles required expensive and specialized equipment. This study has tested the capabilities and potential of sensors found in common smart mobile phones.
With the increasing growth of cities and the lack of space, indoor parking garages have become a vital component for any transportation system. Since emergencies will occur sooner or later, preplanning is necessary especially for indoor parking as their closed environments and limited access rank them among the most dangerous places in case of disaster. Therefore, preparing a successful emergency evacuation plan with the aim of reducing evacuation time as much as possible is necessity to prevent any chaos among drivers. Providing guidance to distressed drivers trying to exit a facility as quickly as possible requires an accurate condition of the current state of the facility (number of vehicles in the facility, location/distribution of these vehicles, current constraints, as well as their temporal and spatial evolution). An initial general traffic plan that best fits the situation can be selected from a pre-defined evacuation plan library. A particular traffic plan is obtained and implemented in the field by the traffic planners and managers at the facility. Control and guidance techniques contribute significantly to traffic management. Besides, this study offered a sensitivity analysis and a dynamic calibration framework, which can be used for computationally expensive models. This procedure can lead to advanced emergency decisions and can guide emergency planners through the most effective, efficient strategies and actions that should be taken into consideration to have a successful evacuation plan in case a disaster occurred.
This research presented an integrated methodological framework for parking management under constraints. The needs for data, systems, technologies and software were identified and the overall methodology could be implemented for:
- Enhancing precision positioning indoors
- Increased reliability of simulation models for large-scale parking lots
- Interactive control strategies for drivers’ guidance
- Efficient management of emergency cases
- New prospects for innovative business plans