•BEV adoption analysis with 1.6 millon people for 3years travel behavior data is developed.•Human travel mode detection model and travel habit clustering model are proposed.•Diverse consumption ...attitudes are taken into consideration.•A novel weighted vehicle adoption potential metric is introduced.•The detailed advises for policy target response are presented.
With the pressing demand of climate change mitigation, the rapid technological development and market adoption of battery electric vehicles are imperative. However, the diverse consumption attitudes and their interactions, which are rarely considered, can significantly affect the adoption potential of battery electric vehicles. On the basis of three years of global positioning system data of 1.6 million people, we estimated the travel and adoption demands of battery electric vehicles in Japan considering diverse consumption attitudes. Under the current construction conditions for public charging systems and charging technologies, the adoption potential of battery electric vehicles may not be as promising as previously expected, and the government still faces great pressure to respond to the market share target. Given the current level of battery technology, technical and policy improvements such as fast charging, reducing the production cost, perfecting the public charging infrastructure, and increasing purchasing subsidies were found to be more effective than improving the battery capacity at increasing the adoption potential of battery electric vehicles.
•Proposing a framework to assess the emission reduction potential of ride-sharing from empirical mobility data.•Proposing a deep learning-based method that shows higher performance than traditional ...simulation method.•Over 1 million historical individual GPS records in Tokyo area are employed for analysis.•We found that averagely 26.97% of traveling distance can be saved and maximally 84.52% of reduction of CO2 emission.
Spreading green and low-consumption transportation methods is becoming an urgent priority. Ride-sharing, which refers to the sharing ofcarjourneys so that more than one person travel in a car, and prevents the need for others to drive to a location themselves, is a critical solution to this issue. Before being introduced into one place, it needs a potential analysis. However, current studies did this kind of analysis based on home and work locations or social ties between people, which is not precise and straight enough. Few pieces of research departed from real mobility data, but uses time-consuming methodology. In this paper, we proposed an analysis framework to bridge this gap. We chose the case study of Tokyo area with over 1 million GPS travel records and trained a deep learning model to find out this potential. From the computation result, on average, nearly 26.97% of travel distance could be saved by ride-sharing, which told us that there is a significant similarity in the travel pattern of people in Tokyo and there is considerable potential of ride-sharing. Moreover, if half of the original public transit riders in our study case adopt ride-sharing, the quantity of CO2 is estimated to be reduced by 84.52%; if all of the original public transit riders in our study case adopt ride-sharing, 83.56% of CO2 emission reduction can be expected with a rebound effect because of increase of participants from public transit. Ride-sharing can not only improve the air quality of these center business districts but also alleviate some city problems like traffic congestion. We believe the analysis of the potential of ride-sharing can provide insight into the decision making of ride-sharing service providers and decision-makers.
Recently, object detectors based on deep learning have become widely used for vehicle detection and contributed to drastic improvement in performance measures. However, deep learning requires much ...training data, and detection performance notably degrades when the target area of vehicle detection (the target domain) is different from the training data (the source domain). To address this problem, we propose an unsupervised domain adaptation (DA) method that does not require labeled training data, and thus can maintain detection performance in the target domain at a low cost. We applied Correlation alignment (CORAL) DA and adversarial DA to our region-based vehicle detector and improved the detection accuracy by over 10% in the target domain. We further improved adversarial DA by utilizing the reconstruction loss to facilitate learning semantic features. Our proposed method achieved slightly better performance than the accuracy achieved with the labeled training data of the target domain. We demonstrated that our improved DA method could achieve almost the same level of accuracy at a lower cost than non-DA methods with a sufficient amount of labeled training data of the target domain.
Recently, deep learning techniques have had a practical role in vehicle detection. While much effort has been spent on applying deep learning to vehicle detection, the effective use of training data ...has not been thoroughly studied, although it has great potential for improving training results, especially in cases where the training data are sparse. In this paper, we proposed using hard example mining (HEM) in the training process of a convolutional neural network (CNN) for vehicle detection in aerial images. We applied HEM to stochastic gradient descent (SGD) to choose the most informative training data by calculating the loss values in each batch and employing the examples with the largest losses. We picked 100 out of both 500 and 1000 examples for training in one iteration, and we tested different ratios of positive to negative examples in the training data to evaluate how the balance of positive and negative examples would affect the performance. In any case, our method always outperformed the plain SGD. The experimental results for images from New York showed improved performance over a CNN trained in plain SGD where the F1 score of our method was 0.02 higher.
The prediction of human mobility can facilitate resolving many kinds of urban problems, such as reducing traffic congestion, and promote commercial activities, such as targeted advertising. However, ...the requisite personal GPS data face privacy issues. Related organizations can only collect limited data and they experience difficulties in sharing them. These data are in "isolated islands" and cannot collectively contribute to improving the performance of applications. Thus, the method of federated learning (FL) can be adopted, in which multiple entities collaborate to train a collective model with their raw data stored locally and, therefore, not exchanged or transferred. However, to predict long-term human mobility, the performance and practicality would be impaired if only some models were simply combined with FL, due to the irregularity and complexity of long-term mobility data. Therefore, we explored the optimized construction method based on the high-efficient gradient-boosting decision tree (GBDT) model with FL and propose the novel federated voting (FedVoting) mechanism, which aggregates the ensemble of differential privacy (DP)-protected GBDTs by the multiple training, cross-validation and voting processes to generate the optimal model and can achieve both good performance and privacy protection. The experiments show the great accuracy in long-term predictions of special event attendance and point-of-interest visits. Compared with training the model independently for each silo (organization) and state-of-art baselines, the FedVoting method achieves a significant accuracy improvement, almost comparable to the centralized training, at a negligible expense of privacy exposure.
Customized Bus is a new mode of Internet-supported public transportation. It is regarded as one of the major strategies to reduce the usage of private cars and mitigate greenhouse gas emissions from ...road traffic. For a customized bus system, a dynamic bus line planning system based on the demand can largely improve the performance and promote the public acceptance of customized bus service. This paper introduces a method to generate planning suggestions for bus lines and stops based on massive demand data. A link network is generated from the input to represent the sharing route of the demand. With community detection, the link network is segmented into communities with similar travel routes. By examining the core-peripheral structure and matching the core part of communities with the road network, the customized bus lines are generated. Boarding and alighting hotspots are identified as the suggestion for customized bus stops. The methodology is tested by using mobile phone data in Tokyo. With the input of one-day sample, the algorithm can generate the result in approximately 1 min and extract 29 bus lines. According to the shape and spatial location of the bus lines, three types of bus lines serving different travel patterns are classified: radiation type lines, ring-type lines, and suburban lines. Analyzing the emission reduction potential of the extracted bus lines manifestes that bus line planning of the proposed method has the potential to relieve emission pressure on urban expressways and to reduce approximately 13% of road traffic emission.
Display omitted
•A data-driven approach is proposed for the systematic planning of customized bus.•1.4 million car trajectories are used to simulate dynamic input for the method.•Dynamic bus lines are generated from spatial distribution of sharing trajectories.•Suggestion of bus stops are extracted from boarding and alighting hotspots.•Potential demand and emission reduction of customized buses is analyzed.
Pedestrian trajectory prediction under crowded circumstances is a challenging problem owing to human interaction and the complexity of the trajectory pattern. Various methods have been proposed for ...solving this problem, ranging from traditional Bayesian analysis to Social Force model and deep learning methods. However, most existing models heavily depend on specific scenarios because the trajectory model is constructed in absolute coordinates even though the motion trajectory as well as human interaction are in relative motion. In this study, a novel trajectory prediction model is proposed to capture the relative motion of pedestrians in extremely crowded scenarios. Trajectory sequences and human interaction are first represented with relative motion and then integrated to our model to predict pedestrians' trajectories. The proposed model is based on Long Short Term Memory (LSTM) structure and consists of an encoder and a decoder which are trained by truncated back propagation. In addition, an anisotropic neighborhood setting is proposed instead of traditional neighborhood analysis. The proposed approach is validated using trajectory data acquired at an extremely crowded train station in Tokyo, Japan. The trajectory prediction experiments demonstrated that the proposed method outperforms existing methods and is stable for predictions of varying length even when the model is trained with a controlled short trajectory sequence.
The transportation sector has become the leading and most-rapidly growing contributor to greenhouse gas emissions. Promoting low-carbon travel mode is critical to alleviate this issue. As a new ...travel mode, online ride-hailing (such as Didi Chuxing and Uber), is becoming increasingly popular in cities around the world. However, there is still no comparative analysis on fuel consumption and emissions: does online ride-hailing have a distinct fuel consumption and emissions pattern with traditional taxis? In this study, we use one month global positioning system dataset and orders dataset, averagely covering around 7 thousand taxis with 0.3 million trips and 23 thousand Didi Chuxing Express vehicles with 0.1 million trips per day in Chengdu, China to answer this question. Empirical results show that taxi trips associate with longer idle distance and shorter delivery distance than Didi trips. Didi trips’ average idle velocity is apparently smaller than their delivery velocity. Online ride-hailing mode is concluded to contribute to these difference: after dropping off previous passengers, Didi drivers usually park their cars until being dispatched new orders and then drive directly to pick up passengers rather than search circuitously. Fuel consumption and carbon monoxide, nitrogen oxides, hydrocarbon emissions per passenger-on kilometer of taxi trips are found to be about 1.36, 1.45, 1.36 and 1.44 times that of Didi trips, respectively. Additionally, only taxi drivers with good performance have the ability to reduce fuel consumption and emissions; while most Didi drivers can perform well on fuel consumption saving and emissions reduction. Finally, several feasible policies are suggested for improving and upgrading the traditional taxi business. Our study provides convincing evidence for understanding the advantage of online ride-hailing mode in reduction fuel consumption and emissions sourced from empty cruising, so as to support better traffic policy making and the promotion on low-carbon travel mode.
The outbreak of coronavirus disease (COVID-19) has swept across more than 180 countries and territories since late January 2020. As a worldwide emergency response, governments have implemented ...various measures and policies, such as self-quarantine, travel restrictions, work from home, and regional lockdown, to control the spread of the epidemic. These countermeasures seek to restrict human mobility because COVID-19 is a highly contagious disease that is spread by human-to-human transmission. Medical experts and policymakers have expressed the urgency to effectively evaluate the outcome of human restriction policies with the aid of big data and information technology. Thus, based on big human mobility data and city POI data, an interactive visual analytics system called Epidemic Mobility (EpiMob) was designed in this study. The system interactively simulates the changes in human mobility and infection status in response to the implementation of a certain restriction policy or a combination of policies (e.g., regional lockdown, telecommuting, screening). Users can conveniently designate the spatial and temporal ranges for different mobility restriction policies. Then, the results reflecting the infection situation under different policies are dynamically displayed and can be flexibly compared and analyzed in depth. Multiple case studies consisting of interviews with domain experts were conducted in the largest metropolitan area of Japan (i.e., Greater Tokyo Area) to demonstrate that the system can provide insight into the effects of different human mobility restriction policies for epidemic control, through measurements and comparisons.
As a fundamental parameter of the electric grid, obtaining spatial electric load distribution is the premise and basis for numerous studies. As a public, world-wide, and spatialized dataset, ...NPP/VIIRS night-light satellite image has been long used for socio-economic information estimation, including electric consumption, while little attention has been given to the electric load estimation. Additionally, most of the previous studies were performed at a large spatial scale, which could not reflect the electric information inner a city. Therefore, this paper proposes a method to estimate electric load density at a township-level spatial scale based on NPP/VIIRS night-light satellite data. Firstly, we reveal the different fitting relationships between EC (Electric Consumption)-NLS (Night-Light Sum) and EL (Electric Load)-NLI (Night-Light Intensity). Then, we validated the spatial-scale’s influence on the estimation accuracy by experiment via generating a series of simulated datasets. After working out the super-resolution night-light image with the SRCNN (Super-Resolution Convolutional Neural Network) algorithm, we established a finer spatial estimation model. By taking a monthly data of Shanghai as a case study, we validate the model we established. The result shows that estimating electric load at township-level based on night-light satellite data is feasible, and the SRCNN algorithm can improve the performance.
•Electric load estimation is conducted based on night-light satellite data.•Fitting relationships between electric Load and night-light intensity is revealed.•The spatial-scale’s influence on the estimation model fitting is validated.•A finer electric load estimation method based on the SRCNN algorithm is conducted.