21 Comments


  1. Huhh,
    Life is strange.
    I have solved this problem earlier with a different method, but did not really find a good application for it.
    They say “meter scale precision” is possible.
    I think I could do millimeter scale if needed.
    Just put a simple camera on the heel and a very small device creating the motion vector based on differences on the photographs taken (let say) in every half second. Like your laser mouse. Really simple technology…


  2. what about when the smoke from the fire reduces the cameras effectivness 🙂


  3. Yeah, this is what I said, I did not locate a nice application for it. I thought about darkness and snow. Both is no problem for a camera. Smoke can be. But, is there such a fire reducing visibility to below 10 cm? Then surely nobody would be able to move around.


  4. There is no way you could maintain position with a heel-based camera (at least not with today’s technology).


  5. Have a look on your optical mouse. It works perfectly even on a single white sheet of paper.
    The technology is very simple.
    Anyways, I thought we can fix my equipment on the bottom of cars and have it running throughout its lifetime. A new owner can read its data to see the car’s history. Average, maximum speed, distances, G forces, …


  6. Just because it’s easy to understand, doesn’t make it a simple technology. Yes, on an even and consistent surface where the sensor will never move more than a few inches per second and is always at a fixed distance, it’s “easy” to do.

    If you tried to put this on a mechanism that was moving variable distances from the surface at high speeds you’d run into a ton of problems. You’d need higher camera resolution, nearly real-time focusing of the camera on the surface you’re walking on, a ton of processing power to do all the image processing and a way to keep it clean so it works after mud splashes on it.

    But if you have some magical way to fix all of these problems, by all means, design it and put it on the market.


  7. The truth must be in between two of us.
    You said impossible, I said possible.
    The technology of creating motion vectors from two camera shots is readily available (hardware is available). I did not say I can fix all other problems like mud.
    Gyroscope, Accelerometer and Magnetometer also need lots of energy for above guys.


  8. “I have solved this problem earlier with a different method”
    “I think I could do millimeter scale if needed.”
    “I thought we can fix my equipment on the bottom of cars”

    Based on these statements, you’re implying that you have this technology ready to go. There are obviously a ton of applications for it, why have you not began to sell it?


  9. Tell me tons of applications.
    I (we) can do the stuff for you.
    I work for a company, doing very interesting image processing software, hardware and optics for it.
    We have discussed the technology and it shows that very high level of accuracy can be achieved. Just to give you an example we have been discussing a projection system for subway wall advertisement projection where the car position and speed would have been measured by this method. Calculated the things and that showed it was within the range. (Due to other things we dropped the project…)

    But I do not use GPS, so do not really no where it is not working properly in cities, or maybe walking people should ever buy GPS-like stuff which can direct them based on their initial position fixed when started walking in the morning.


  10. know not “no”*


  11. Cool, I had thought about doing this a while ago but never actually tried it. If you can measure the relative movement and you know the absolute starting position, then you can easily determine the absolute position.

    Sticking it in a boot is a great idea, I hadn’t thought of that. It gives you much more accuracy.


  12. Interesting idea, but similar “relative position tracking” projects have been developed in the past and they all suffer from some serious problems for long(er) term use. The main issue is sensor noise. Recall, that in order to track position with an accelerometer, one has to integrate twice (once for velocity). This means that even a small amount of noise can be magnified to outrageously incorrect positions. The way individuals alleviate this problem is to use other sensors (such as the gyroscope and magnetometer in this project) to “correct” the data, often using a Kalman filter. However, the problem still remains that over time, sensor drift will cause their accuracy to drift from 1 meter to 10 meters to …

    Also, why would they use Bluetooth for the data transmission? Bluetooth is notoriously inadequate for long distance communication through buildings (i.e., not line-of-sight).


  13. They’re only using Bluetooth to go from the module to their first responder radios; those look like they handle comms back to the truck.

Comments are closed.