conference paper
Phd Dissertation
Estimating and forecasting travel demand have been a popular study topic among transportation researchers; however the research needs to pursue new direction with the advent of data from the potential availability of newer types of data previously not envisaged. In this dissertation, the author reviews previous studies on this topic and develops approaches for two aspects of travel demand analysis in the transportation network: A newer OD estimation method and a household activity-based demand modeling framework. First, a trip-based dynamic OD estimation model is developed. Several previous studies on OD trip table estimation focused on a static problem and many recent dynamic OD estimation methods also have not sufficiently proved their practical applicability. In order to overcome the shortcomings, this dissertation introduces supplementary information (i.e., vehicle trajectory data) to a dynamic OD estimation model. However, the trip-based approach has certain well-known shortcomings. OD estimation results can not give satisfactory solutions for forecasting purposes, and the estimated OD table only contains materialized trips, which implies that no latent travel demand is included in the table. Therefore, the estimated OD table does not have sufficient information for identifying the real travel demand pattern and it is not so useful for transportation planning works. Contrarily, a standard four-step model has a better capability for explaining a travel demand pattern. However, when we load the OD trip table calculated by the four-step model, we might see some discrepancies between simulated traffic patterns and the ground truth. The discrepancies can come from various factors such as insufficient network capacities and unexplained influencing factors. When the discrepancy is caused by insufficient network capacities, then it can be solved by an iterative adjusting procedure. Using the ground truth such as link traffic counts, it might be updated correctly. However, if the discrepancies come from incapability of the four-step model, then we should look for a new approach. The capability of the four-step model already has been criticized continuously by numerous activity researchers because a trip-based approach does not correctly consider the real motivation of travel. To overcome these drawbacks, the second item of fucud in the dissertation is in developing a dynamic agent-based household activity and travel demand simulation model framework named DYNAHAP. The framework calculates a demand pattern in terms of activity chains generated by synthetic families. A traffic simulator then executes the activity chains, and finally an aggregated dynamic traffic pattern is generated. In order to calibrate DYNAHAP, huge activity data should be gathered. Such tasks had been regarded very difficult or even nearly impossible before, but with the development of data collecting technologies, currently we have several ways for collecting the activity chains of individuals. Like vehicle trajectory data, sample activity chains collected from personal communication devices such as PDA (Personal digital assistant) could be used for DYNAHAP calibration. Some numerical test results also will be given for proving the performance of the developed models. In last chapter, some important issues for future study are also discussed.
research report
conference paper
California, where transportation accounts for over half of ozone precursors and particulate matter emissions, as well as nearly 40 percent of greenhouse gas emissions, has adopted the ambitious goal of reducing petroleum use in transportation by 50 percent by 2030. One of the proposed strategies to achieve this goal is to increase the share of alternative fuel vehicles (AFVs). Current incentives to foster the adoption of AFVs by households include single-occupant use of High Occupancy Vehicle (HOV) lanes and parking privileges. Although popular, the effectiveness of these incentives is still controversial. To shed some light on this question, this paper analyzes data from the 2012 California Household Travel Survey using a generalized structural equation model that accounts for residential self-selection, household demographic characteristics, and environmentalism. The results show that, households who live closer to freeways with HOV lanes, work closer to parking lots with AFV privileges, and have pro-environmental views are more likely to own AFVs. However, these households are also likely to drive slightly more than if they had conventional vehicles, which is not surprising since AFVs have a lower variable cost compared to conventional vehicles. On average, a household who lives 1 mile closer to HOV lanes will drive an additional 0.27 mile per month for work or school (0.88 mile if a parking facility with AFV privileges is 1 mile closer to work.) These unintended effects suggest that alternative measures such as pricing vehicle miles traveled (VMT) should be considered if policymakers decide to also reduce VMT.
book/book chapter
research report
Data available from the inaugural year (1978-1979) of the UMTA Section 15 requirement are examined for three purposes: 1) to assess reliability of the data; 2) develop a small set of performance indicators; and 3) to produce a classification of bus systems based upon inherent characteristics. To assess the reliability of the transit data, econometric models based upon previous data sets were replicated. Improvements in data collection are recommended. Development of performance indicators was accomplished using factor analysis. Forty-eight performance measures were grouped into nine performance dimensions. A standardized value on each performance indicator was calculated for each transit property. Transit properties were ranked by their sum on each of the nine performance indicators. Several methods were tested for clustering transit systems into peer groups. The most satisfying clusters were based upon four variables: two representing size–active buses and annual vehicle miles–and two representing nature of operations — average speed and peak to base ratio. Eight groups of transit properties were identified and described in terms of the four variables. The performance of 198 properties on the nine indicators is listed by group. Properties are identified by code number not by name. Despite the inadequacies found in the data collected in the inaugural year, methods were developed which can help managers and administrators data to improve transit management.
published journal article
Autonomous vehicles are cyber-physical systems that combine embedded computing and deep learning with physical systems to perceive the world, predict future states, and safely control the vehicle through changing environments. The ability of an autonomous vehicle to accurately predict the motion of other road users across a wide range of diverse scenarios is critical for both motion planning and safety. However, existing motion prediction methods do not explicitly model contextual information about the environment, which can cause significant variations in performance across diverse driving scenarios. To address this limitation, we propose CASTNet: a dynamic, context-aware approach for motion prediction that (i) identifies the current driving context using a spatio-temporal model, (ii) adapts an ensemble of motion prediction models to fit the current context, and (iii) applies novel trajectory fusion methods to combine predictions output by the ensemble. This approach enables CASTNet to improve robustness by minimizing motion prediction error across diverse driving scenarios. CASTNet is highly modular and can be used with various existing image processing backbones and motion predictors. We demonstrate how CASTNet can improve both CNN-based and graph-learning-based motion prediction approaches and conduct ablation studies on the performance, latency, and model size for various ensemble architecture choices. In addition, we propose and evaluate several attention-based spatio-temporal models for context identification and ensemble selection. We also propose a modular trajectory fusion algorithm that effectively filters, clusters, and fuses the predicted trajectories output by the ensemble. On the nuScenes dataset, our approach demonstrates more robust and consistent performance across diverse, real-world driving contexts than state-of-the-art techniques.
Phd Dissertation
Advanced traffic management is a cost-effective option to reduce total delay, fuel consumption and air pollution in urban networks. Nevertheless, Adaptive Signal Control, the most advanced scheme for real-time traffic responsive operations, is still not widely used due to inadequate sensor systems and the deficiencies in the control algorithms. A novel traffic data system was recently proposed at UC Irvine named the “Persistent Traffic Cookies” (PTC) system, in which the routes traveled by the vehicles are recorded onboard and read using short-range wireless communication among vehicles and roadside devices. An advantage of this system is that there is no requirement of massive central databases and data processing of all possible vehicles in the network. The accumulated travel data is distributed across vehicles. The trip behavior inferred in the day-by-day data is used to predict individual paths and aggregated across vehicles for traffic prediction in dynamic network traffic control. This research develops traffic control schemes that use path-based data systems like PTC. Initially, methods are presented to generate the required path-based input variables such as turning flows and travel times. Two main aspects are addressed. One is a systematic approach to define spatial boundaries of subnetworks for area-control using observed traffic dynamics, the path flow between signalized intersections being used as the criterion for control dependency. The second focus is to provide network-level signal optimization, based on a decentralized control scheme yielding indirect signal coordination optimized for delay with no explicit bandwidth maximization. The local optimization uses a Dynamic Programming approach using the predicted arrival flows modeled via link traffic platoon dispersion. Optimal signal indications are found for small time steps (currently 5 seconds) within the control horizon, essentially resulting in a “cycle-less” operation. A modified rolling horizon scheme is applied, incorporating a proper calculation of the salvage cost of left-over queue after the horizon. Signal coordination is indirectly achieved and the feedback among signal decisions lead to an iterative approach. The schemes are evaluated with a microscopic simulation study of a real-world network. The results showed that the scheme reduces the total delays in the network in comparison to the Actuated Signal Control already installed in the network. It is also seen that the modified rolling horizon method with salvage cost considerations performs better than the more conventional methods.
working paper
The vehicle choice model developed here is one component in a micro simulation demand forecasting system being designed to produce annual forecasts of new and used vehicle demand by vehicle type and geographic area in California. The system will also forecast annual vehicle miles traveled for all vehicles and recharging demand by time of day for electric vehicles. The choice model specification differs from past studies by directly modeling vehicle transactions rather than vehicle holdings. The model is calibrated using stated preference data from a new study of 4747 urban California households. These results are potentially useful to public transportation and energy agencies in their evaluation of alternatives to current gasoline-powered vehicles. The findings are also useful to manufacturers faced with designing and marketing alternative-fuel vehicles as well as to utility companies who need to develop long-run demand side management planning strategies.
working paper |
Jan 1999
book/book chapter |
Oct 1998
working paper |
Sep 1994
working paper |
Dec 1994
conference paper
Complete and accurate traffic information is becoming more and more available with the advance in transportation surveillance technology. Especially, vehicle classification information can contribute to many transportation related fields such as road pavement management, estimation of polluted emission etc. In previous sections and chapters, it was shown that vehicle signature is function of vehicle type and traffic conditions. By exploiting this concept, the algorithm development in vehicle classification is investigated and corresponding results are presented.
