Peng Hao | University of California, Riverside (original) (raw)

Papers by Peng Hao

This paper presented a deep reinforcement learning method named Double Deep Q-networks to design ... more This paper presented a deep reinforcement learning method named Double Deep Q-networks to design an end-to-end vision-based adaptive cruise control (ACC) system. A simulation environment of a highway scene was set up in Unity, which is a game engine that provided both physical models of vehicles and feature data for training and testing. Well-designed reward functions associated with the following distance and throttle/brake force were implemented in the reinforcement learning model for both internal combustion engine (ICE) vehicles and electric vehicles (EV) to perform adaptive cruise control. The gap statistics and total energy consumption are evaluated for different vehicle types to explore the relationship between reward functions and powertrain characteristics. Compared with the traditional radar-based ACC systems or human-in-the-loop simulation, the proposed vision-based ACC system can generate either a better gap regulated trajectory or a smoother speed trajectory depending o...

1 2 The public raises concerns about the exposure to particulate matter (PM) which has been stron... more 1 2 The public raises concerns about the exposure to particulate matter (PM) which has been strongly 3 associated with illness and mortality. However, most of the studies rely on the measurements from 4 stationary monitoring sites which cannot capture the actual PM exposure for those people in or 5 near the source. In this study, we first set up a comprehensive mobile monitoring platform to 6 measure both PM concentration and traffic conditions on some major highways in Southern 7 California. Then, we developed an integrated database to fuse different data sources and to 8 facilitate the investigation of relationship between traffic conditions and highway PM 9 concentration. Using the fused datasets and combining with Emission FACtor (EMFAC) model, 10 contour plots based on estimated PM emissions were generated with the overlay of particle 11 concentration measurements. Analyses of the results indicate that there are numerous particle 12 concentration peaks cause by traffic congesti...

Author(s): Hao, Peng; Wei, Zhensong; Bai, Zhengwei; Barth, Matthew J. | Abstract: The Eco-Approac... more Author(s): Hao, Peng; Wei, Zhensong; Bai, Zhengwei; Barth, Matthew J. | Abstract: The Eco-Approach and Departure (EAD) application has been proved to be environmentally efficient for a Connected and Automated Vehicles (CAVs) system. In the real-world traffic, traffic conditions and signal timings are usually dynamic and uncertain due to mixed vehicle types, various driving behaviors and limited sensing range, which is challenging in EAD development. This research proposes an adaptive strategy for connected eco-driving towards a signalized intersection under real world conditions. Stochastic graph models are built to link the vehicle and external (e.g., traffic, signal) data and dynamic programing is applied to identify the optimal speed for each vehicle-state efficiently. From energy perspective, adaptive strategy using traffic data could double the effective sensor range in eco-driving. A hybrid reinforcement learning framework is also developed for EAD in mixed traffic condition u...

Author(s): Ye, Fei; Hao, Peng; Qi, Xuewei; Wu, Guoyuan; Boriboonsomsin, Kanok; Barth, Matthew

The Transportation Research Board (TRB) 99st Annual Meeting, Washington D.C., January 12-16, 2020

2019 IEEE Intelligent Transportation Systems Conference (ITSC), Oct 1, 2019

SAE Technical Paper Series, Apr 14, 2020

E co-Approach and Departure (EAD) has been considered as a promising eco-driving strategy for veh... more E co-Approach and Departure (EAD) has been considered as a promising eco-driving strategy for vehicles traveling in an urban environment, where information such as signal phase and timing (SPaT) and geometric intersection description is well utilized to guide vehicles passing through intersections in the most energy-efficient manner. Previous studies formulated the optimal trajectory planning problem as finding the shortest path on a graphical model. While this method is effective in terms of energy saving, its computation efficiency can be further enhanced by adopting machine learning techniques. In this paper, we propose an innovative deep learning-based queue-aware eco-approach and departure (DLQ-EAD) system for a plug-in hybrid electric bus (PHEB), which is able to provide an online optimal trajectory for the vehicle considering both the downstream traffic condition (i.e. traffic lights, queues) and the vehicle powertrain efficiency. Based on optimal solutions obtained from the graph-based trajectory planning algorithm (GTPA), a deep neural network (DNN) is developed to learn the optimal vehicle speed for the next time step given its current state. It is demonstrated that the trained DNN can provide energyefficient trajectories with high computational efficiency and high f lexibility adopting to dynamic changes in the surrounding environment. To address the impact of downstream traffic, a queue prediction model is further developed using data from radars and connected vehicles (CVs), as well as signal timing data from SPaT messages. A comprehensive simulation study in the microscopic traffic modeling software PTV VISSIM shows that the proposed DLQ-EAD can achieve 18.7%-24.0% energy efficiency improvements for a single PHEB on various traffic congestion levels. The proposed queue prediction model can be of practical significance even at low penetration rates of CVs. Specifically, additional energy savings of 2.0%-8.2% can be further achieved with 20% vehicles in the network.

2017 IEEE Intelligent Vehicles Symposium (IV), Jun 1, 2017

technology has become increasingly popular. As an example, Cooperative Adaptive Cruise Control (C... more technology has become increasingly popular. As an example, Cooperative Adaptive Cruise Control (CACC) systems are of high interest, allowing CAVs to communicate and cooperate with each other to form platoons, where one vehicle follows another with a predefined spacing or time gap. Although numerous studies have been conducted on CACC systems, very few have examined the protocols from the perspective of environmental sustainability, not to mention from a platoonwide consideration. In this study, we propose a vehicle-to-vehicle (V2V) communication based Eco-CACC system, aiming to minimize the platoon-wide energy consumption and pollutant emissions at different stages of the CACC operation. A full spectrum of environmentally-friendly CACC maneuvers are explored and the associated protocols are developed, including sequence determination, gap closing and opening, platoon cruising with gap regulation, and platoon joining and splitting. Simulation studies of different scenarios are conducted using MATLAB/Simulink. Compared to an existing CACC system, the proposed one can achieve additional 2% energy savings and additional 17% pollutant emissions reductions during the platoon joining scenario.

2015 International Conference on Connected Vehicles and Expo (ICCVE), Oct 1, 2015

Prior studies have shown that tangible environmental benefits can be gained by communicating the ... more Prior studies have shown that tangible environmental benefits can be gained by communicating the driver with the signal phase and timing (SPaT) information of the upcoming traffic signal with fixed time control. However, similar applications to actuated signals pose a significant challenge due to their randomness to some extent caused by vehicle actuation. Based on the framework previously developed by the authors, a preliminary real-world testing has been conducted to evaluate the system performance in terms of energy savings and emissions reduction. Four scenarios that covers most of traffic and signal conditions are evaluated. For each scenario, we vary the entry time and speed to thoroughly investigate the performance of the developed Eco-Approach and Departure (EAD) system. It turns out that the EAD system saves 5%-10% energy for high entry speed, and 7%-26% energy for low entry speed. That results is compatible with the simulation results and validate the previously developed EAD framework.

IEEE Transactions on Intelligent Transportation Systems

Transportation Research Part A: Policy and Practice

Most special-use freeway lanes, such as High Occupancy Vehicle (HOV) lanes, have traditionally be... more Most special-use freeway lanes, such as High Occupancy Vehicle (HOV) lanes, have traditionally been designed with either limited access or continuous access control from the adjacent generalpurposed (GP) lanes. Studies have shown the advantages and disadvantages of each design in terms of safety, mobility, environment, and enforcement, among other factors. With a focus on improving the operational performance of HOV facilities, this paper proposes a new design called partially limited access control where the continuous access is mostly designated along the freeway to achieve higher travel speed while buffers between the HOV lane(s) and the adjacent GP lanes are strategically placed on selected freeway segments to accommodate higher throughput on those segments. The placement of buffers primarily aims to reduce the impact of HOV cross-weave flow on the capacity of GP lanes. In this research paper, a methodology for determining the location and length of buffers in the partially limited access control has been developed. A case study is performed along a 13-mile section of HOV facility on SR-210 E in Southern California, which is coded and evaluated in traffic microsimulation. The results show that the partially limited access control increases the throughput (represented by total vehicle miles traveled or VMT) and decreases the delay (represented by total vehicle hours traveled or VHT) of the freeway as compared with either the limited access or continuous access control. As a result, the overall efficiency (represented by average travel speed calculated as VMT/VHT) of the freeway with partially limited access HOV facility is 21% and 6% higher than that of the freeway with limited access and continuous access HOV facility, respectively, under the baseline traffic demand.

We introduce in this paper the concept of vehicle indices in a cycle at a signalized intersection... more We introduce in this paper the concept of vehicle indices in a cycle at a signalized intersection which are the positions of vehicles in the departure process of the cycle. We show that vehicle indices are closely related to the vehicle arrival and the departure processes at the intersection. Based on vehicle indices and sample travel times collected from mobile sensors, a three-layer Bayesian Network model is constructed to describe the stochastic intersection traffic flow by capturing the relationship of vehicle indices, and the arrival and departure processes. The non-homogeneous Poisson process and log-normal distributions are used respectively to model the stochastic arrival and departure processes. The methods of parameter learning and vehicle index inference are presented based on the observed intersection travel times. Simplification to the methods is discussed to reduce the computational effort of parameter learning and vehicle index estimation. The model is tested using data from NGSIM, a field test, and simulation with reasonable results.

Transportation Research Part B: Methodological, 2015

Queue length is one of the key measures in assessing arterial performances. Under heavy congestio... more Queue length is one of the key measures in assessing arterial performances. Under heavy congestion, queues are difficult to estimate from either fixed-location sensors (such as loop detectors) or mobile sensors since they may exceed the region of detection, which is defined as long queue in the literature. While the long queue problem has been successfully addressed in the past using fixed-location sensors, whether this can be done using mobile traffic sensors remains unclear. In this paper, a queue length estimation method is proposed to solve this long queue problem using short vehicle trajectories obtained from mobile sensors. The method contains vehicle trajectory reconstruction models to estimate the missing deceleration or acceleration process of a vehicle. Long queue estimation models are then developed using the reconstructed vehicle trajectories. The proposed method can provide estimates of the queue profile and the maximum queue length of a cycle. The method is tested in a field experiment with reasonable results.

Probe vehicles that measure position and speed have emerged as a promising tool for traffic data ... more Probe vehicles that measure position and speed have emerged as a promising tool for traffic data collection and performance measurement, but the sampling rates of most probe vehicle sensor data available today are low (ranging from 10 to 60 s per sample), and the data coverage is limited. Therefore, it is challenging to accurately estimate the vehicle dynamic states in both space and time based on these sparse mobile sensor data. In this paper, a stochastic model is proposed to estimate the second-by-second vehicle speed trajectories by examining all possible sequences of modal activities (i.e., acceleration, deceleration, cruising, and idling) between consecutive data points from sparse position and speed measurements. The likelihood of occurrence of each sequential pattern is first quantified by mode-specific a priori distributions. The vehicle dynamic state probability is then formulated as the product of probabilities for multiple independent events. Therefore, a detailed vehicle speed trajectory can be reconstructed using the optimal modal activity sequence, which maximizes the likelihood. The proposed model is calibrated and validated using the Next-Generation SIMulation dataset. The results show the substantial improvements on the accuracy of estimated vehicle trajectories compared with a baseline method based on linear interpolation. The proposed model is applied to a large-scale vehicle activity dataset to demonstrate the estimation of hourly traffic delay variation.