The High Cost of Embedded Computers

Every morning I like to go for a nice 10-mile bicycle ride. There are two factors that determine how quickly I will complete my ride. The first is if there is any wind. The second is how many times I get stopped because of traffic. Constantly being stopped because of traffic lights has me despising them more and more. It also has me thinking about the large computers that run them and why they have to be so big and expensive.

This got me thinking about how home and office computers have become so powerful and relatively inexpensive while the functionality required for a traffic light has remained unchanged for decades yet the cost hasn't matched normal computer pricing trends. Furthermore it seems that computers embedded in everyday objects such as cars are equally expensive. When you buy a new car the salesperson will justify the always increasing price with the complex computer inside. The funny thing is that automotive and traffic light computers are really simple and could be handled by the same computer chips found in ancient computers from the 1980's.

If you have never seen a traffic light computer, it is the large box located on one of the corners of the intersection it guards. They are generally about 4-feet tall, 3-feet wide, and 2-feet deep. That's a big box for something that can be run by a less powerful chip than I have in my watch. Now I can understand the size when lights required so much power to remain visible during daylight hours. Now we have ultra-low-power LED lights that don't need much power at all. So what is the big box for? My guess is it has to withstand a car running into it. It probably also goes towards justifying the $70K required for each intersection. Think about it though, if you could run a traffic light off the same computer found in a watch, you could put it in a much smaller box on one of the poles and save taxpayers a significant amount of money.

I know there are a lot of embedded computers that have dropped in price as one would expect. After all we have a lot of toys available that wouldn't be possible without such inexpensive electronics. I just wish those cost savings could be passed along to more things. Things I find I am paying more and more for that I honestly don't really want nor like. 

How Smart Traffic Lights Work

One of the advantages of taking Electrical Engineering classes in college and having a degree in the subject is that I learned how a number of devices in our society work. The other day I saw someone who sent their children to jump up and down on a sensor used to detect if a car is overhead. In this case they were trying to open an arm for a parking lot but it is the same technology for traffic lights. Unfortunately this adult erroneously believed that traffic sensors are triggered by weight. I thought the same thing growing up. Imagine my surprise when I attended my Electromagnetic Physics class and the professor informed us that it is just a metal detector buried under the asphalt?

Before college I tried like those kids to get my bicycle to trigger traffic lights to turn green for me. After doing a number of bunny-hops with no luck, I would sheepishly walk over and push the pedestrian button. After college I learned a trick and can now trigger traffic signals to turn green for me about 80% of the time. The bike I currently use to commute has an aluminum frame. While aluminum is a great conductor of electricity, it isn't magnetic and that is what a metal detector is looking for. Iron is magnetic and also rather heavy so they try to use as little as possible of it in bicycles. There is one place they can't avoid it though and that is the bike chain. I have discovered that if you ride so your chain is directly over the sensor, there is enough iron in it to trigger the light. This may take a bit of trial and error for some lights but I have learned where all of the sensors are on my 3-mile ride to and from work each day.

When I saw those kids jumping up and down on the car sensor, I politely asked them to move and rode over the sensor. My trick worked and they entered the parking lot without any trouble. It is a simple trick but looks like magic to young kids.

Smart Stoplights

There are three stoplights on my way to and from work each day. Considering I mostly use a bike path and and am only on the road for less than two miles, that seems to be a bit much. All of the lights are considered "smart lights" in that they have sensors in the pavement to detect when a car is present. Only one of the three can be triggered by a bicycle and so I have to hit the pedestrian button on the other two.

For some reason, a lot of people think that smart lights are triggered by the weight of the car. That isn't true. Instead there is a metal detector built into street. When you car passes over the detector (basically a piece of wire wound around a toilet paper tube), a signal is sent to the smart light's computer. In California, the system actually works and lights change accordingly. In Utah, they have some sort of algorithm that was written by some idiot that never got full credit for homework assignments in college. There is a bit of a delay that can be as long as a minute. This is meant to keep traffic flowing in high-congestion area. Unfortunately it seems like a huge waste of my time at 11:30 at night when I am coming home from the airport.

I would like to propose a new for of smart light. Instead of using a metal detector embedded in the street, why don't we use the camera mounted on top of the stoplight? After all, we have facial recognition software that does a pretty good job. Shouldn't we also be able to create automobile and bicycle recognition software? One of the problems with the metal detector method is that you have to be right over the sensor. Using a camera, you could see the cars approaching and determine that you want to keep traffic flowing as opposed to instantly switching lights. In the case where there are no cars coming because it is 11:30 at night, light changes would be much quicker.

Hopefully someone is thinking about this. After all, most stoplights already have cameras on them to monitor traffic flow and accidents. I'll bet we can even improve everyone's gas mileage as we will spend less time idling and more time at speed. Most importantly, I won't have to get off my bike to hit the pedestrian button to cross the street.