Building a better mouse trap, with video surveillance

My mom has been using this style of drowning mouse trap at the cottages the last few years. The advantage of this trap is that it can catch many mice in one trap.

The trap consists of a coat hanger wire across the bucket, with an aluminium can on the wire. The can has peanut butter smeared onto it. As the mouse tries to get at the bait, the can spins, and the mouse falls into the bucket.

I started using this style of trap in my workshop in the country. But I was curious to see how the mice actually used it. When I tried monitoring it last spring there weren't any mice in the shop. The mouse problem peaks in the fall when the mice look for an escape from the cold, so this fall (2015) I tried again.

I wanted to observe how the mice actually got caught in the trap by monitoring it with my Raspberry Pi and camera module and my imgcomp software.

In the mean time I had sealed my workshop against mice a bit better, so I set up my experiment in a shed, running an extension cord to power my setup.

The monitoring setup consists of a Raspberry Pi model B+, mounted on my Raspberry pi and camera module holder. The camera is aimed at the trap. The pi is running some software that I wrote for surveillance. It takes three pictures per second, and if it sees any changes from one to the next, it saves it.

I also have a lamp aimed at the setup with my. Mice do come out even when it's light so long as it's quiet.

I didn't put any water in the bucket this time. I didn't want to show mice getting killed on YouTube. The mice can just barely jump out of the bucket. I figure that way, the mouse gets a second chance, and I get more observations.

Checking the setup with my imgcomp software. As soon as my hand enters the field of view, my software detects motion and records it.
A mouse visitor. The mouse ate some of the bait off the can without falling in. Partly, it used its tail as a balance off the bucket's rim. It never tried climbing onto the unstable can. I only had one mouse visit during the first week of observation.


Next I moved the can closer to the edge of the bucket to make it more enticing. But this only resulted in the mouse systematically eating all the bait within reach, never trusting the can enough to get fully on it. It even deliberately rotated the can to get at more bait.

Clever mouse!

My next thought was to give the mouse a semi-stable perch in the form of a "weighted" can. So I put a few small stones in one can, which I put next to another freely spinning can.

This approach worked much better. The weighted can was stable enough for the mouse to climb onto without it flipping over. The mouse used the weighted can to try to get onto the freely spinning one, though it never went far enough onto it to get dumped. But as the mouse tried to get the bait towards the edges of the weighted can, it reached as far as it could without losing grip. But the can would shift a bit from the weight of the mouse, and that was enough for the mouse to lose its grip. This happened a number of times over the course of a week. The mouse was able to jump back out of the bucket each time, though sometimes it took more than an hour of trying.

The weighted can clearly works and is better at catching mice than the un-weighted can.

But I had another idea: a tipping ramp. As the mouse gets nearer the front, the ramp tips down, dumping the mouse into the bucket.

I tried making a ramp out of some 30 mm wide metal strapping, bending the strapping over several times to form a counterweight on the back. But this ramp was heavy. With all it inertia, it would tip slowly, giving the mouse time to scurry back to safety.

I wanted a ramp that would dump spontaneously without warning.

So here is the design I came up with.

It's a very light wooden ramp, with a screw in the back for a counterweight. The top of the ramp has some plastic on it from some packaging so that it's too smooth for the mouse to claw into.

The counterweight is only heavy enough to bring the ramp back to horizontal. The ramp is mostly held up by a very small magnet in the back, which pulls down on a piece of paper clip wire on the back of the ramp.

The magnet is stuck to a washer by it's own magnetism. That way, I can tweak its position.

The idea is that the magnet holds the ramp well enough for the mouse to get onto the ramp, but once its force is exceeded and it starts to tip, the holding power of the magnet drops rapidly with distance. So basically, the ramp appears stable until it gets too much load, and then it dumps spontaneously.

This ramp worked exceedingly well. The same mouse would get on it time and time again! Very satisfying to watch!

Also interesting to watch the mouse get out of the bucket. Unfortunately, at 3 frames per second, it was hard to see how the jumping happened. The mouse is able to jump to nearly the edge of the bucket.

These two frames are very revealing about how the mouse got out. In the left frame, you can see a mouse paw on the slight ledge of the bucket, and in the next frame, the mouse is on the bucket's edge, further to the left.

So the mouse jumped in a more or less helical path. This meant the mouse had some momentum towards the wall, which allowed it to get a slight boost with one of its paws off the narrow beveled ledge in the bucket, and that was enough to make it up to the rim. Clever mouse!

Over time, the mouse wised-up to the trap, grabbing hold of the edges of the ramp as it went down. Though sometimes it still fell in while trying to climb back up.

A possible improvement would be to make the ramp edges curve down with a larger radius so the mouse couldn't grab onto them. Though, realistically, this is only a problem once the mouse gets practice with the trap. If there was water in the trap, the mouse would be long dead before it had a chance to figure it out.

But the ramp-based trap is very tricky to make and tweak just right, so I wouldn't recommend you build one like that. The un-weighted can on a coat hanger wire works (caught quite a few mice with it in the past). And weighting the can makes it semi-stable, which makes the mouse more likely to get on it and fall off.

See also:

Imgcomp -- The software I used for this video.

Rat trap

I later tried the same design of trap, but scaled up about 1.6x to try to catch a rat, but with much less success. Video below.


To my woodworking website.