Nice project, and will probably work great for short-term studies, but you'll want to improve the battery life and work hard on weatherproofing if it's to be used in the field for anything other than temporary projects.
I used to work for a company (http://www.trafx.net/) that specialized in making a device just like this, but using a passive infrared sensor instead of the ultrasonic sensor. The number one field problem is moisture damage to the electronics. Condensation, UV failure of seals, ice expansion through seals, etc. The list goes on. Based on the moss on the trees in the photo, it looks like a pretty wet environment, so you'll want to double-triple-quadruple seal that case and sensors before you think about leaving it outside for more than a month or two.
For comparison, we had some customers that had had our units in continuous use for nearly 10 years. Battery life on 3xAA alkaline was about 60 months using the IR sensor. The Arduino+ultrasonic sensor is a pretty power hungry combination for a battery-operated unit that might need to operate for extended periods. 5 days is a pretty short study period, a lot of the useful statistics that land management organizations want to see will span multiple years.
thanks for feedback. you are right, the battery life is poor.
the problem with the battery life is not the arduino itself - a bare atmega328 (the hearth of arduino) will live for several months from a 3AA battery.
the real power consumption comes from the ultrasonic sensor - I tried to completely cut it off from power between measurements, however the ultrasonic circuit is not recovering properly from these on&offs (I was looking for help on arduino forums but without response: http://forum.arduino.cc/index.php?topic=216079.msg1589123#ms...)
This may be a dumb question, but isn't this more of a "movement detection counter" rather than a "people counter"? Nothing sets apart an animal vs a human triggering this, right? I was imagining a camera with CV processing the images or something.
The ultrasonic transducer probably does a pretty good job of warding off animals. Nothing quite like disturbing the wildlife while counting how many people are disturbing the wildlife.
hello Alex. thanks for remark, you are right. However my "ultimate" goal is to count people passing a back-country skiing track, so I decided to emphasise the human-counting use case.
I'd look seriously at the TRAFx counters I referred to in my other comment. They'll do exactly what you want, they're used to count backcountry skiers by quite a few organizations, they work great in winter down to at least -40C.
There are some non-trivial design problems that need to be overcome in outdoor environments. The ultrasonic sensor may not be a good choice for winter counting, frost buildup on the sensor surface will likely greatly affect performance, plus it's a bit of a power hog.
Don't forget that battery capacity/max current can be greatly affected by cold temperatures as well.
I used to work for a company (http://www.trafx.net/) that specialized in making a device just like this, but using a passive infrared sensor instead of the ultrasonic sensor. The number one field problem is moisture damage to the electronics. Condensation, UV failure of seals, ice expansion through seals, etc. The list goes on. Based on the moss on the trees in the photo, it looks like a pretty wet environment, so you'll want to double-triple-quadruple seal that case and sensors before you think about leaving it outside for more than a month or two.
For comparison, we had some customers that had had our units in continuous use for nearly 10 years. Battery life on 3xAA alkaline was about 60 months using the IR sensor. The Arduino+ultrasonic sensor is a pretty power hungry combination for a battery-operated unit that might need to operate for extended periods. 5 days is a pretty short study period, a lot of the useful statistics that land management organizations want to see will span multiple years.