Car-following Control, i.e. human drivers, and Bilateral Cruise Control (demonstrated by 20 Gizmo Garden Robots)

  • Prof. Berthold K. P. Horn will tell you the story of bilateral cruise control


  • Bilateral cruise control solves traffic problems (e.g. stop-and-go and collisions caused by car-following control). Roomlight is used to switch the modes between car-following control and bilateral cruise control. LED is used to indicate car-following control mode (Yellow) and bilateral cruise control mode (blue LED)


  • These two videos will help you understand the above demo: 1. Details of the "cars"£º An brief introduction of the basic components and functions of the "cars" (i.e. Gizimo-garden robots) used for demonstration; 2. Demo with commentary£º Prof. Horn explains more details about the above demonstration.

  • Build a Gizmo-Garden robot, i.e. a simplified prototype of "car", by yourself. 1. The two documents manual 1 and manual 2 from Sharp provide the characteristics of the distance sensors, and some description of how they work. In summary, they operate by triangulation using an IR emitter and a linear array sensor. This makes them largely immune to the reflectivity of the surface. 2. Arduino MEGA2560 is used to implement the control system of the "car" (i.e. Gizmo-Garden robot). 3. Parallax Robotics Shield Kit is used to implement the mechanical system of the "car" (i.e. Gizmo-Garden robot).

  • In summary, traffic problem can be solved by the cars themselves if bilateral cruise control is widely used.

  • These experimental results are summarized in our recent paper: "Bilateral cruise control: smooth traffic flow from simple local interactions." (submitted to Science Robotics)

    • o What is bilateral cruise control?

  • All these videos are to help you understand 1). what bilateral cruise control is, 2). why bilateral cruise control can solve the traffic problem and 3). why bilateral cruise control is implementable.
  • video 1£º An brief introduction of bilateral cruise control.
  • video 2£º Cool simulations of car-following model and bilateral cruise control.
  • video 3£º The simulation result of our Java-based on-line simulater.
  • video 4£º The difference between bilateral cruise control (BCC) and traditional platoning model (Note that there is no wireless communication or preset desired speed in this demonstration. All information comes from the sensors). Prof. Horn also talks about today's adaptive cruise control (ACC) model.

    • o What is Gizmo Garden Robot?

  • The robots used for demonstration are built by Gizimo gardeners. This video shows you the big idea of Gizimo garden project and Gizimo garden robots. These videos also help you understand 1). How Gizmo Garden Robot works, 2). why Gizmo Garden Robot is a simplified prototype of self-driving car.
  • video 1£º Basic components in Gizimo garden robot.
  • video 2£º Infrared sensors are used for longitute control.
  • video 3£º Reflective photointerrupters are used for steering.
  • video 4£º Cars run on the road now. The Gizimo garden robots will be used to demonstrate the traffic.

    • o Specify Gizmo Garden Robots for the demonstrations

  • All these videos are to help you understand 1). What the purpose of each component of Gizmo Garden Robut for our experiments, 2). what information we can obtain from the dash pannel, 3) these robots real work, 4). the validation of these robots for implementing car-following (CF) control and bilateral cruise control (BCC) model, 5). 5) what information tells by the LED lights in our later experimental demonstrations, and 6). the validation of swith the CF mode and BCC mode by controling the room light. These videos should be convincing about the results shown in the later experimental demonstrations.
  • video 1£º An brief introduction of basic components of the robots, and their use for our experimental demonstrations.
  • video 2£º The munus And how to set the parameters for the experiments.
  • video 3£º The various driving modes for the experiments.
  • video 4£º The robot moves under car-following (CF) control.
  • video 5£º The robot moves under bilateral cruise control (BCC).
  • video 6£º Using room light sensor to choose the modes of CF or BCC.
  • video 7£º Using the LED to show whether the robot is speeding up or slow down.
  • video 8£º From BCC mode to CF mode by changing roomlight (demonstration).mp4.

    • o The experiments

  • All these videos are to show you 1). the "stop-and-go" instability caused by car-following control, 2). The power of bilateral cruise control solving the problem so efficiently.
  • video 1£º The "stop-and-go" traffic jam (and collisions) in nowadays traffic under car-following control.
  • video 2£º The demonstration of car-following and bilateral cruise control with Prof. Horn's commentary.
  • video 3£º We see the similar result in the demo using less robot (i.e. 13 cars) and shorter road.
  • video 4£º A longer video (11 min. 38 sec.) of the demonstration of car-following and bilateral cruise control. Fig.1 and Fig. 2 shows the (forward) space and speed of Car 2 during car-following control (black curves) and bilateral cruise control (red curves) periods.

    • Fig.1: The measured space by car 2 (coresponding to video 4). black curve is the measurement during the car-following control period. Red black curve is the measurement during the BCM period. Under CFM, obvious ¡°stop-and-go¡± pattern appear, i.e. large space and small space alternating. Under bilateral cruise control, traffic flow instabilities are suppressed effectively. The space is kept about 200 mm. No small space. Failure of detection due to the turning of cars on the corner and noise in the distance measurements cause the fluctuation in space measurement. However, the motion of the traffic is still pretty smooth. Thus, bilateral cruise control can be used in reality robustly.


    Fig.2: The speed of car 2 (coresponding to video 4). black curve is the measurement during the CFM period. Red black curve is the measurement during the BCM period. Under CFM, obvious ¡°stop-and-go¡± pattern appear, i.e. high speed and very low speed alternating. Under BCM, traffic flow instabilities are suppressed effectively. The speed is kept about 90 mm/s. No very low speed.


  • Fig.3 and Fig. 4 shows the (forward) space and speed of Car 13 during CFM (black curves) and BCM (red curves) periods.

    • Fig.3: The measured space by car 13 (coresponding to video 4). black curve is the measurement during the CFM period. Red black curve is the measurement during the BCM period. Under CFM, obvious ¡°stop-and-go¡± pattern appear, i.e. large space and small space alternating. Under BCM, traffic flow instabilities are suppressed effectively. The space is kept about 200 mm. No small space. Failure of detection due to the turning of cars on the corner and noise in the distance measurements cause the fluctuation in space measurement. However, the motion of the traffic is still pretty smooth. Thus, BCM can be used in reality robustly.


    Fig.4: The speed of car 13 (coresponding to video 4). black curve is the measurement during the CFM period. Red black curve is the measurement during the BCM period. Under CFM, obvious ¡°stop-and-go¡± pattern appear, i.e. high speed and very low speed alternating. Under BCM, traffic flow instabilities are suppressed effectively. The speed is kept about 90 mm/s. No very low speed.


    A poem "usiing bilateral control model"

    Smartly driving is pretty simple.
    Just try to stay in the middle!
    No traffic jams or other trouble,
    because traffic flow will be stable!
    It's not platoon model
    controlling your cars by lead people!
    Even the benefit might be double,
    my car won't answer to other people!
    Using bilateral control model!
    Let your cars themselves handle!
    Remember to stay in the middle!
    We'll be far away from trouble!
    That's bilateral control model
    making traffic flow stable!