Connected and automated vehicles have recently been in the spotlight, partially led by the U.S. Department of Transportation. Major public initiatives have explored the roles of different detection and communication technologies in future transportation systems. These technologies, however, also can be utilized to collect detailed traffic data on both vehicular and nonvehicular flows in multimodal environments, with and without connectivity and automation. In this workshop, researchers present their latest efforts using different technologies—lidar, dedicated short-range communications, Bluetooth, GPS tracking and smartphones, simulators, and virtual reality and gaming—to collect traffic data.
This presentation offers the challenges associated with establishing the Australian Integrated Multimodal EcoSystem (AIMES) in Melbourne, Australia. The AIMES is an urban laboratory supporting the implementation, validation and testing of emerging connected transport technologies and associated operational scenarios. The Test Bed provides a large scale complex urban environment in which testing can be undertaken in a live environment. The AIMES itself will include a network of diverse intelligent and distributed sensors technologies. We will use the Test Bed to study, test and deploy a variety of transport connectivity technologies. This work describes how the use of different technologies from Transport Management Platform through to roadside sensors combined with the latest advances in data processing will be brought together to facilitate a test environment to enable the evaluation of tomorrows technology in a real live multi-modal environment in a busy city center.
Traveler and driver behavior data are needed in a changing connected multi-modal transportation environment. Traveler behavior is the outcome of decisions associated with departure time, destination, mode and route choices in a dynamic commute environment characterized by a wealth of information and an increased number of travel options. Driver behavior should be also captured and modeled given the different levels of acceptance of the automation levels being introduced in an era where driverless cars are seen as part of the future. There have been different data collection platforms suggested to collect such traveler and driver behavior data including games, simulators, instrumented/probe vehicles, videos/cameras …etc. However, until today, such data collection platforms have been characteristic of a given mode of travel or a given controlled environment/experiment. In this presentation, we offer online platforms and cell-phones as a data collection platform for high-resolution trajectory data and traveler behavior data and for multiple modes of transportation (including bikes, personal vehicles, walking, personal vehicles …etc). A sample interactive dashboard has been utilized where data was collected on traveler behavior in Northern Virginia. After presenting the corresponding results, we will show how such dashboard has been integrated as a mobile application (app.). Such integration allows utilizing the accelerometer and GPS data from the cell-phone censors to reconstruct trip characteristics as well as trajectories for different modes of transportation. Technical details on the data collected in the USA and in Spain and the corresponding required specifications will be shared with the audience.
Recently, there is considerable interest in developing simulator or virtual world for set-driving vehicles. We present AutonoVi-Sim, a novel high-fidelity simulation platform for testing autonomous driving algorithms. AutonoVi-Sim is a collection of high-level extensible modules which allows for the rapid development and testing of vehicle configurations, and facilitates construction of complex road networks. Autonovi-Sim supports multiple vehicles with unique steering or acceleration limits, as well as unique tire parameters and overall vehicle dynamics profiles. Engineers can specify the specific vehicle sensor systems and vary time of day and weather conditions to gain insight into how conditions affect the performance of a particular algorithm. In addition, AutonoVi-Sim supports navigation for non-vehicle traffic participants such as cyclists and pedestrians, allowing engineers to specify routes for these actors, or to create scripted scenarios which place the vehicle in dangerous reactive situations. AutonoVi-Sim also facilitates data analysis, allowing for capturing video from the vehicle's perspective, exporting sensor data such as relative positions of other traffic participants, camera data for a specific sensor, and detection and classification results. Thus, AutonoVi-Sim allows for the rapid prototyping, development and testing of autonomous driving algorithms under varying vehicle, road, traffic, and weather conditions. We highlight its performance in traffic and driving scenarios.
Connectivity and automation are expected to transform the current driving environment by providing safer and more efficient driving experience. Unveiling the full potential of these technologies requires careful planning, which itself requires a thorough understanding of these technologies and their impact on the transportation system. The current state-of-the-practice in designing and operating connected automated vehicles (CAVs) relies heavily on data and deep learning based systems. While such information (including collected data and model structures) is critical to assessing the impacts of connectivity and automation on transportation systems, it is currently unavailable to researchers (i.e., companies consider data as potential commodity). Accordingly, in this presentation, the efforts to build CAVs for data collection and model development are offered based on a recent collaborative initiative between Texas A&M University and George Washington University. The initiative consists of developing two automated Kia Souls in College Station, TX and Washington D.C. Details on the corresponding technical requirements and limitations are presented. Sample online datasets collected from these CAVs are shared with the audience followed by an introduction to our traffic simulator developed based on this dataset.
With the rapid development of the autonomous vehicle industry, it is recognized that autonomous vehicles may have significant impact on traffic flow dynamics. Many efforts have been devoted to study the impact of autonomous vehicles on traffic flow stability, throughput and environmental footprints in a mixed traffic flow of both autonomous vehicles and regular human-driven vehicles. However, most of these studies are based on modelling and simulations. Only very recently, an experimental study has been reported to investigate traffic flow on a circuit track with length 260 meters (Stern et al., 2017, arXiv1705.01693). It has been shown that an autonomous vehicle is able to dampen stop-and-go waves.
We perform an experimental study on traffic flow dynamics in a 21 car-platoon on the track at Chang’ An University, China. The length of the track is about 2.4 km. While 20 cars are regular human-driven vehicles, one autonomous vehicle is put in the middle of the platoon. The leading car moves with several different constant speeds, and other cars follow each other without overtaking. We collect the speed and trajectory of each car with high-precision GPS device. We study the impact of the autonomous vehicle on the development of traffic oscillation.
This talk will discuss the collection and use of new data for traffic research. Urban mobility requires more and more attention these days. Instead of focusing on a single mode, urban mobility requires multi-modality. Assessing performance of urban traffic as well as its prediction and control require accurate empirical data. We present how with various data collection techniques we can observe traffic and learn behavior. We aim to collect data on travelers and traffic conditions, ranging from local observations (using loop detectors) to network observations for OD tables and routes. We will introduce RADD, the Researchlab Automated Driving Delft, with its focus on automated driving and Wifi and GPS sensors to observe pedestrians and cyclists.
In addition, we will show applications of data from smart phones; in particular, the position of the phone can be found by triangulation, Wifi sensors and GPS. Google collects data of this type, when the users explicitly agree to share. While these data might not be very precise, accurate and even the penetration rate is unknown, we still can derive insights of the traffic stream. As example, we analyze urban-scale traffic dynamics and estimate the queue length on an controlled freeway on-ramp.