3D Robotics

uBlox releases a dead reckoning chip


This isn't a GPS module, but rather a chip that compliments GPS modules by using dead reckoning to estimate position during GPS drop-outs. APM and Pixhawk do this already in software, but it would be interesting to see if a stand-alone hardware solution is useful for UAV purposes (this first chip is aimed at automotive use).

From Gizmodo:

Swiss company u-blox just devised a chip that keeps you on course when the satellites drop out, using the most old-fashioned of navigation techniques: dead reckoning.

The u-blox 3D Dead Reckoning chip uses a bevy of accelerometers, gyroscopes, and speed sensors to calculate a car's exact location, no GPS signal needed. By measuring the direction, speed and distance the car has traveled since the last GPS check-in, the chip can tell if the car's still on the intended route, or if it's veered off into lost-land. U-blox's demonstration video, narrated by a charming Walter Cronkite imitator, explains in more detail:

The 3D Dead Reckoning chip isn't packed in to any currently-available GPS systems, and it's not really clear whether the company intends to sell these things to equipment manufacturers or directly to consumers—though the press release indicates it shouldn't be too hard to install yourself.

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • Gary McCray
    As it works in all three dimensions I'd guess that the filtering (kalman?) is working under some assuptions like that the car will not travel straight up or roll in some axes... that makes the calculations all the more reliable... and that's why it would probably not work well in aerospace applications since those kinds of movements are pretty common whether we like it or not :P

  • Developer

    Gary, that is true. That chip/solution would kill it at ground part of the next Sparkfun AVC. With odometer + dead reckoning + good GPS coverage, I imagine you could pretty much run flat out for the entire track.

  • From reading their specs, it looks like this is actually already embedded in a GPS chip, not standalone:


    Although this chip seems dependent on recording actual odometer based ground travel it could work well in some rover or robotic applications.

    And if we can make better use of technologies to accurately record ground track (flow camera for instance) it might be useful to us as well.

    From the video it is clear they are functioning in all 3 dimensions (they mention pitch, but they would have to do roll as well to compensate for tilt in the parking garage up and down ramps to achieve the kind of accuracy shown.

    And they would also have to use accels to compensate for gyro error also around those same up and down ramps.

  • unfortunately, I dont't think it would add much to any UAV aplications... Since cars have virtually two degrees of freedom less than any aircraft dead reckoning in their case is far easier... and they use precise odmeters so it is virtually only a matter of calculating direction of travel, task which current ardupilot code covers pretty damn well:)
    Yet any such development is a step in a right direction...

  • Developer

    You already find similar systems in newer cars with built in satellite navigation. By using wheel ticks (odometer) to measure distance traveled, those systems can be very accurate. Here in Norway there are many long tunnels on the roads, and I traveled on of those just a couple of days ago. The tunnel was 7.5KM long with slopes and road position was a accurate all the way trough with no jump at the end when GPS was reestablished.

  • Being a Swiss resident, I am often impressed with their contribution to the electronics of UAV's.
  • Developer

    Hi Nikola,

    The APM code has multiple dead-reckoning systems depending on the sensors available. If you have a compass and airspeed sensor in a fixed wing aircraft (ie. in APM:Plane) then the code used is in libraries/AP_AHRS/AP_AHRS_DCM.cpp

    For a multicopter it uses the inertial nav code in libraries/AP_InertialNav

    We will also be shortly merging Pauls Kalman filter code into the APM code that has very nice dead-reckoning capabilities which should be much more accurate than the current code. That will be in libraries/AP_NavEKF

    Cheers, Tridge

  • Moderator

    Hi Chris ,

    for our application could be interest evaluate the quality of imu sensor gyro and acc , but don't have the odometer available on uav so is not good for fly application . I think that could be interesting for evaluate rover indoor application .

  • Where do I find the dead-reckoning code in the APM code base?

  • This looks cool, it will be very interesting to see how it stacks up against our solution.

This reply was deleted.