Real-time Signal Control and Traffic Stability / Improved Models for Managed Lanes Operations
1 like187 views
The document discusses Stephen D. Boyles' research on real-time signal control and traffic stability, particularly for managed lane operations. It highlights the adaptation of backpressure concepts for managing traffic at reservation-based intersections and introduces improved models for dynamic toll lanes. Future work includes testing alternative methods and calibrating models with field data to optimize throughput and revenue in traffic management.
Real-time Signal Control and Traffic Stability / Improved Models for Managed Lanes Operations
- 1. Real-time signal control and traffic stability Improved models for managed lanes operations Stephen D. Boyles Associate Professor The University of Texas at Austin April 11, 2018 DSTOP updates Boyles
- 2. PROJECT 125: REAL-TIME SIGNAL CONTROL AND TRAFFIC STABILITY
- 3. The reservation-based intersection Connected and automated vehicles provide new opportunities for improving intersection throughput. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 4. Earlier DSTOP research showed that improving capacity locally may not improve throughput globally. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 5. This project explores better ways to prioritize vehicles at reservation-based intersections. Conversations with DSTOP colleagues raised the possibility of backpressure-based intersection priority. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 6. How to adapt the backpressure concept for roadway networks? Physical queues: Roadway links have a finite (and often binding) “buffer” for storing vehicles, which bounds pressure terms. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 7. How to adapt the backpressure concept for roadway networks? Physical queues: Roadway links have a finite (and often binding) “buffer” for storing vehicles, which bounds pressure terms. Solution: “Chain” queues that spill back by recursively adding pressure terms for links upstream. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 8. How to adapt the backpressure concept for roadway networks? Physical queues: Roadway links have a finite (and often binding) “buffer” for storing vehicles, which bounds pressure terms. Solution: “Chain” queues that spill back by recursively adding pressure terms for links upstream. Route choice: Vehicles can choose routes independently and selfishly DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 9. How to adapt the backpressure concept for roadway networks? Physical queues: Roadway links have a finite (and often binding) “buffer” for storing vehicles, which bounds pressure terms. Solution: “Chain” queues that spill back by recursively adding pressure terms for links upstream. Route choice: Vehicles can choose routes independently and selfishly Solution: Add equilibrium principle and iterate with updated routes. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 10. Simulation environment DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 11. Granting reservations according to the backpressure principle reduced delays significantly beyond earlier (FCFS) protocols. DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 12. Future work Test alternative way to address finite buffer (nonlinear transformation of pressure term) DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 13. Future work Test alternative way to address finite buffer (nonlinear transformation of pressure term) Compare with P0 policy developed by M. J. Smith DSTOP updates Project 125: Real-time signal control and traffic stability Boyles
- 14. PROJECT 140: IMPROVED MODELS FOR MANAGED LANE OPERATIONS
- 15. Managed lanes Dynamic toll lanes present new opportunities and challenges for traffic management. DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 16. Earlier research is largely confined to single-entrance, single-exit facilities. DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 17. In facilities with multiple entrances and exits, the number of paths through the corridor grows exponentially with the number of access points. DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 18. DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 19. By reformulating the route choice model to operate at diverge node, we can represent all possible paths with a polynomial set of variables. DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 20. We used approximate dynamic programming to identify toll policies which maximize throughput (public facility) and maximize revenue (private facility). DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 21. DESE network Policy Revenue ($) TSTT (hrs) Throughput (veh) Selected policy VFA1 $217.35 39.71 1586 Selected policy VFA2 $253.53 39.50 1596 Selected policy VFA3 $487.92 41.42 1533 Selected policy VFA4 $41.49 33.88 1639 Density based heuristic $27.87 36.81 1561 Ratio based heuristic $106.65 35.5 1586 Myopic revenue policy $31.44 36.11 1576 LBJ network Policy Revenue ($) TSTT (hrs) Throughput (veh) Selected policy VFA1 $554.94 45.80 1185 Selected policy VFA2 $593.78 45.61 1191 Selected policy VFA3 $419.85 43.20 1210 Selected policy VFA4 $88.92 40.69 1266 Density based heuristic $79.92 40.79 1266 Ratio based heuristic $115.29 41.36 1240 Myopic revenue policy $328.05 41.27 1234 DSTOP updates Project 140: Improved models for managed lane operations Boyles
- 22. Future work Use field data to calibrate/validate model. Explore methods for estimating value-of-time distributions from data. Further explore “jam-and-harvest” phenomenon DSTOP updates Project 140: Improved models for managed lane operations Boyles