Detector-Free Optimization of Traffic Signal Offsets with Connected-Vehicle Data
Christopher Day, Iowa State UniversityShow Abstract
Howell Li, Purdue University
Lucy Richardson, Kimley-Horn and Associates, Inc.
James Howard, Indiana Department of Transportation
Tom Platte, Indiana Department of Transportation
James Sturdevant, Indiana Department of Transportation
Darcy Bullock, Purdue University
It has recently been shown that signal offset optimization is feasible using vehicle trajectory data at low levels of market penetration. This study performs offset optimization on two corridors using this type of data. Six weeks of trajectory splines were processed for two corridors including 25 signalized intersections, in order to create vehicle arrival profiles, using a proposed procedure called “virtual detection.” After processing and filtering the data, penetration rates between 0.09–0.80% were observed, varying by approach. The arrival profiles were statistically compared against those measured with physical detectors, with the majority of the approaches showing statistically significant goodness-of-fit at a 90% confidence level. Finally, the virtual detection arrival profiles were used to optimize offsets, and compared against a solution derived from physical detector arrival profiles. The results demonstrate that virtual detection can produce good quality offsets with current market penetration rates of probe data. The study also includes a sensitivity analysis to the sample period, which shows that two weeks of data may be sufficient for data collection at current penetration rates.
Adaptive Coordination Based on Connected-Vehicle Technology
Byungho Beak, University of ArizonaShow Abstract
Larry Head, University of Arizona
Yiheng Feng, University of Michigan, Ann Arbor
This paper presents a methodology that integrates adaptive coordination with adaptive signal control in a connected vehicle environment. The model consists of two levels of optimization. At the intersection level, an adaptive control algorithm allocates the optimal green time to each phase in real time using dynamic programming considering coordination constraints. At the corridor level a mixed integer linear program is formulated based on data from the intersection level to optimize offsets along the corridor. After the algorithm solve the optimization problem, the optimized offsets are sent back to the intersection level algorithm to update the coordination constraints. The model is compared with actuated-coordinated signal control using VISSIM simulation.
Spatiotemporal Intersection Control in a Connected and Automated Vehicle Environment
Yiheng Feng, University of Michigan, Ann ArborShow Abstract
Chunhui Yu, Tongji University
Henry Liu, University of Michigan, Ann Arbor
This paper proposes a joint control framework of traffic signals and vehicle trajectories in a connected and automated vehicle environment. The control framework is modeled as a bi-level optimization problem with signal control on the upper level and vehicle trajectory control on the lower level. The signal optimization is modeled as a dynamic programming (DP) problem with the objective to minimize vehicle delay. Optimal control theory is applied to the vehicle trajectory control problem with the objective to minimize fuel consumption and emissions. In order to solve the problem efficiently, the bi-level problem is reformulated as a single-level problem by adopting a simplified objective function which solves the lower level optimal control problem analytically. Simulation results show that the proposed joint control framework is able to reduce both vehicle delay and emissions under a variety of demand levels compared to fixed-time and adaptive signal control when the vehicle trajectories are not optimized. The reduced vehicle delay and CO2 emission can be as much as 24.0% and 13.8%, respectively.
Combining Model-Predictive Intersection Control with Green Light Optimal Speed Advisory in a Connected-Vehicle Environment
Simon Stebbins, University of QueenslandShow Abstract
Mark Hickman, University of Queensland
Jiwon Kim, University of Queensland
Hai Vu, Monash University
In recent years there has been much interest in Connected Vehicle (CV) technology. Vehicle-to-infrastructure (V2I) communication has the potential to reduce delays, stoppage time, fuel usage and emissions, because it allows fine-grained traffic movement data to be shared with greater frequency. Previously, traffic control algorithms have been based on macroscopic, fluid mechanical traffic models, but since V2I communication allows for fine-grained traffic data, a more accurate, microscopic, car-following traffic model will be used instead in this paper.
At an intersection there are essentially two ways to improve traffic conditions – by improving the intersection control schedule, and by modifying vehicle approach trajectories. In order to best utilise traffic data that varies second-by-second, it is proposed that the optimal control schedule that minimises delay can be found via model predictive control (MPC) with suitable state space reduction techniques. In addition, since the control algorithm utilises an underlying microscopic model, entering vehicles’ trajectories can be modified with Green Light Optimal Speed Advisory (GLOSA). This allows drivers to adjust their speed profiles in order to have an efficient approach trajectory. CV technology allows MPC to be integrated with GLOSA, making the best use of this future technology to improve traffic conditions for all motorists.