NTU-NXP Smart Mobility Testbed

  • Real-time vehicle position & speed data collection with RSU(Road-Side Unit) & OBU (onboard unit) within NTU Campus.

  • Remote Traffic Light Control via MQTT server.

Vehicle tracking by using V2X system

  • By making use of NTU’s smart mobility testbed (V2X system), we can collect the real time vehicle data for cars/buses (equipped with OBU) running inside NTU campus.

  • We are dealing with Urban Traffic Signal Scheduling problem, by making use of the vehicle data, we can estimate/calculate the traffic condition (turning ratio, average speed etc.) for each link.

  • The video below demonstrated our vehicle tracking system, check the road/link number changes in the table (upper right corner) when vehicle changes road. Unique ID has been assigned for each road & link within NTU campus.

Data Collection & Analysing – Vehicle Activity Chart

  • We have collected months of vehicle data from the Smart Mobility Testbed.

  • By analyzing historical daily vehicle data, we are able to find the pattern of traffic flow and other useful information.

  • Those historical data can be used to generate training data for us to train neural networks to predict future traffic congestion level.

  • The figure shows number of Active Vehicles in Each Time Interval (10 minutes) For Each Day (Blue – Bus, Yellow – Other Vehicle Types)

Data Collection & Analysing – Vehicle Position Heat Map

  • By putting one day vehicle positions together, we are able to generate the heat map to visualize the most congested regions and vehicle moving area.

Congestion Level Identification

  • For testing purpose we set congestion level to 3 classses:

    • Red (1): speed <= 5km/h

    • Yellow (2): 5km/h < speed <= 10km/h

    • Green (3): speed > 10km/h

  • Congestion level are updated every 10 seconds.

  • The speed means average speed of all vehicles on the same link.


Jurong Area Traffic System Simulation Platform Development in VISSIM

“Green Wave” Generation

Urban Traffic Network Closed-Loop Simulation

Route Suggestion - Show 3 Shortest (Time) Paths

  • Use Google Maps to display 3 shortest paths from position A to B.

    • The shortest paths are calculated based on distance and predicted future road speed profile.

    • the shortest path gives shortest travel time (may not be the shortest distance).

  • Congestion Level Identification

    Pedestrian Traffic Light Scheduling Using PTV Vissim + Viswalk

    Bus (No. 179) Simulation Using PTV Vissim


    1. Y. Zhang and R. Su. Pedestrian phase pattern investigation in a trafficlight scheduling problem for signalized network. 2018 IEEE Conference on Control Technology and Applications. Accepted. 2018.

    2. Y. Zhang, R. Su and Y. Zhang. A macroscopic propagation model for bidirectional pedestrian flows on signalized crosswalks. 56th IEEE Conference on Decision and Control, pp. 6289-6294, 2017.

    3. Y. Zhang, R. Su, Y. Zhang and C. Sun. “Modelling and Traffic Signal Control of Heterogeneous Traffic Systems.” arXiv, 2017.

    4. Y. Zhang, R. Su, K. Gao and Y. Zhang. A pedestrian hopping model and traffic light scheduling for pedestrian-vehicle mixed-flow networks. arXiv, 2017.

    5. Y. Zhang, R. Su, C. Sun and Y. Zhang. Modelling and traffic signal control of a heterogeneous traffic network with signalized and non-signalized intersections. 2017 IEEE Conference on Control Technology and Applications, pp. 1581 - 1586, 2017.

    6. Y. Zhang, R. Su, K. Gao and Y. Zhang. Traffic Light Scheduling for Pedestrians and Vehicles. 2017 IEEE Conference on Control Technology and Applications, pp. 1593 - 1598, 2017.

    7. A. Lentzakis, R. Su and C. Wen. Strategic learning approach to region-based dynamic route guidance. 12th IEEE International Conference on Control & Automation, pp. 842 - 847, 2016.