February 13, 2010
Hacking Traffic Signals
I noticed something funny on my walk tonight. The traffic signals in Pullman are timed funny for pedestrians.
I've always lamented how horrible "downtown" Pullman is for pedestrian traffic. Smash a crosswalk button and you often have to wait for 5 minutes for the lights to change. Most of our intersections have traffic flow sensors on the signals (cheesy little things, I honestly have no idea how they work). I sat at one intersection and tried a few experiments to see how the traffic flow affected crosswalk time.
Experiment 1: Hit the button while traffic is flowing steadily through the intersection. Time to light change (and my 'walk' sign lighting up): 3:30.
Experiment 2: Hit the button while one car is going through the intersection. Little to no traffic following this car (two more cars went through the entire time I waited for this change): 3:30.
Experiment 3: Hit the button while no traffic was flowing through the intersection. There was a ten second gap in traffic. The light changed in 0:15.
Conclusion: The traffic algorithm at this particular intersection is retarded. It must measure traffic flow at the moment the 'walk' button is hit, and decide how long to wait based on that single measurement. As a pedestrian, you should wait until there is a small break in traffic to hit the walk button. You'll probably save time by not hitting the crosswalk button right away...
Given your little experiment I suspect the developers were either stupid, cheap or both. We had to do the hardware timing for asynchronous traffic control, it's a pain, it also requires a more expensive chip. The lazy way is to just wait for the cycle to finish if it detects a car then inject a walk cycle.
Posted by: steve k on March 4, 2010 06:06 AM






by reid
on March 06, 2011
by reid
on November 23, 2009
Time to get out of Pullman, for your sanity!
Posted by: Erik on February 14, 2010 04:55 AM