My experience with the North Paw

For anyone unfamiliar with the North Paw, it is a haptic compass that is worn around the ankle. You can get more info about it here:

I was really excited about the North Paw, as a haptic compass (ideally an internal one, like the South Paw, but we obviously aren't there yet) and therefore a constant and reliable sense of direction has been my dream since I first read about haptic belt experiments in 2004 or 2005.

I got a North Paw and wore it for about three days before deciding that, while the concept is interesting, it is not reliable enough to continue to use in its current form. But, this may partially be due to my own usage pattern. YMMV.

First off, North Paw pros:
  • Surprisingly good battery life. After a full charge, it easily lasted a full day of being out and about and was still going when I took it off after about 12 hours.
  • Despite looking like a house arrest anklet, it's actually pretty discreet and can be easily hidden under a pant leg, providing you aren't wearing skinny jeans.
And, the big con:
  • Usually, for me, it didn't buzz towards North. I only wore it in neighborhoods I'm familiar with while testing, and my city has a pretty well aligned grid, so it was super easy to tell when the North Paw was wrong. Which was most of the time.
I think the main reason the North Paw didn't work for me is that my primary means of transportation is by bike. Coincidentally, the way you calibrate a North Paw is by flipping it over four times. Often, while I was riding my bike, the pedaling motion would apparently trigger this calibration process, causing it to miscalibrate. It is also possible that the metal in the bike gears was throwing off the compass. But, a haptic compass that doesn't work while riding a bike is effectively useless to me, as that is the main time I need to have a good sense of direction. These things make me think that maybe the ankle is a poor location choice for a haptic compass (though probably still better than most other options, which change on more axises even more frequently). The waist might be the right place for such a wearable (and hopefully, one day, implantable).

Another strange thing is that apparently some buildings I frequent (notably the building my office is in, which also manages to act like a giant faraday cage and block out cell phone signals [related? who knows?]) will predictably cause the North Paw to signal a false positive "North". Since my phone compass agrees with this incorrect North, it's clearly the building's fault and not the North Paw's. Still, this makes me question the utility of a perma-compass in general, as its trustworthiness would immediately be suspect whenever you're indoors.


  • What method does it use to tell north? Is it a magnetic north or true? Buildings that are either lots of metal concrete or electricity often wreck havoc on a magnetic compass. As a general rule of thumb when using a basic compass you are told to move away from large boulders and rocks to correct that problem.

    I would try and avoid a magnetic compass in a city setting because of how much magnetic so are playing around in there. I'd be interested to see how it works in a rural area. I may take a look at one.
  • It determines magnetic north only.
  • I was just looking at the page for it and can't believe the price of this thing. It seems so basic.
  • edited March 2016
    Calibration of those can be tricky. It may be possible to improve it with just some firmware changes.
    First off, those sensors don't deliver a true orientation, just some arbitrary positive integer value on 2 axis, some on 3. I'll always recommend the 3 axis ones. Despite the more complex math involved, you get far better results.
    If you have a really good magnetometer, you can one-time calibrate it by spinning/flipping to get the min/max values for your earth magnetic field and be done with it (given you have no other fields influencing). For most devices this does not work. Most drift with temperature so you need to calibrate at least once a day, maybe even multiple times.
    What you can do to improve this: if you take a 3 axis magnetometer and rotate it your recorded values will all, more or less, be on the surface of a sphere.
    For a 2D sensor it'll be anywhere between a sphere/ellipse/line.
    You can describe said sphere mathematically with very few parameters and update them based on the freshly recorded data. You need to filter out the points which are mistakenly recorded by external fields such as magnetized bicycle frames, dc-currents caused fields (often found when traveling in bus/car).
    The filtering get a lot easier once you combine your magnetic sensor data with a 3 axis gyroscope. Those need to be calibrated as well but they are far more precise.  After powerup you calibrate your magnetometer by flip/rotate to get your initial north in a location known to contain a clean earth magnetic field.
    After that, your gyro starts assisting, you can easily spin the thing 10 or 20 times without an additional magnetometer value and get only a few degrees error. You can use the gyro data to predict where on your 3d-value sphere you should be in your compass data. When both data appear to match, you can use the values to keep your sphere parameters updated. If there's a mismatch, just disregard the magnetometer data and keep tracking the orientation with the gyro.
    If you throw in an acelerometer you can make even better choices as you know where your down-vector is and when the device is mostly at rest.
    bonus points: you can even use the magnetometer data to fine tune the gyro calibration over a longer period of time.

    tl;dr: adding more sensory input can overcome the issues experienced and keep good track of north even when doing crazy parkour backflip stunts.

    edit: fixed a mistyped word
  • This is what I was thinking, but much better explained for lack of technical experience with gyroscopes. >~<
  • Easy build, 9DOF sensor and a haptic piece would do the trick. Complexity lies in the code. This is a electrically simple beginner level project if you want to get started with arduino, but the code is pretty tricky to write yourself.
  • Adafruit takes some of the guess-work out of the code. Well, Bosch did that, Adafruit sells it on a board.
  • Nifty, that makes this project a walk in the park if you've got $70 to buy the parts.
  • Right? That's what I thought when I saw the module. Expensive but a months worth of coding/debugging could be better spent.
  • Honestly if I had seen that earlier I would've picked up one on my university research budget. Already using a LSM9DS1, so it's a bit past time for that.

    Will definitely see if I can't fit it in somewhere.
Sign In or Register to comment.