Traffic Light Simulation

1
Traffic Light Simulation

• Lynn Jepsen

2
Introduction and Background

• Try and find the most efficient way to move cars
  through an intersection at different traffic
  densities
• Want to see waiting time and queue length go down
• Green light usage needs to go up
• Many traffic simulations out there, but none that
  I have seen with a learning light

3
Hierarchy
Light Algorithm
Traffic Light
Stores four directions
Simulation
Stores multiple lanes
Direction
Changes lanes for Cars
Lane
Deals with green/red/yellow light
Stores Cars (can't overlap)?
Car
Remembers and Changes speed
Remembers space
4
Simulation
5
Simulation

• Cars on each lane take up a certain number of
  spaces and have a certain speed at each instant
• Speed and space is update every .1 sec along with
  graphics
• Speed up when object in front is getting farther
  away but with a max. accel.
• Slow down for red and most yellow lights
• Use Java to show graphics

6
Light Algorithm (variables)?

• Wants to optimize efficiency
• This is defined as queue length, wait time and
  green light usage
• Independent variables are traffic density, cycle
  length (length of one cycle in intersection) and
  ratio (ratio of green light time in each
  direction)?
• Can NOT change traffic density

7
Light Algorithm

• Uses previous data (past ten cycles)?
• Matching traffic density
• Finds out which cycle had the smallest wait time,
  queue length, and greatest green light usage
• Later averages all three cycles
• Does the same for ratio
• Looking for the best combination
• Some randomness to match randomness of the road
• Should begin to hover around one combo

8
Graphs

• graph all three efficiency variables
• north/south line and an east/west line
• if the algorithm is really optimizing the
  intersection then the two lines are close to each
  other

9
Efficiency

• The intersection can only be so efficient
• If there are just too many cars for the number of
  lanes, then the algorithm will not work as well
  as it would in a lighter traffic density
• This is not an exact science. There are too many
  variables in effect here
• Traffic flow is just too erratic to predict well
• You cant possibly minimize wait time to the
  point where no one waits

10
Results and Conclusions

• The simulation looks realistic
• Light algorithm does cause fewer backups
• Not perfect, but it keeps things under control
  and doesn't allow huge spikes