Open newbsauce opened 9 years ago
Just so I can. I took some time to understand the open-source solutions so I use them as a baseline for sensor fusion solution accuracy.
But for the most accurate solutions I use the EM7180.
I have used the DMP but it is opaque, not particularly stable, and can only do 6 DoF fusion; can't use the magnetometer. This is not very useful IMHO.
The EM7180, like the BNO055 and MAX21100, does "hardware" sensor fusion. Really, just software sensor fusion on the embedded processor. The Max21100 is very good, but MAXIM has gone out of the MEMS sensor business so this is not a viable long-tem sensor fusion solution. The BNO055 is pretty good, but the embedded BMX055 sensor is not, especially the mag. Unfortunately, that's all they offer. The EM7180 has an excellent sensor fusion solution that can take input from a wide variety of 9 DoF sensors. This is what I would recommend if you are not getting what you need from Madgwick or Mahony and don't want to invest the time to reinvent the wheel.
I have lots of writeups on these sensor fusion solutions and how they compare. Look at the wiki on the MPU6050 page.
Kris
-----Original Message----- From: newbsauce [mailto:notifications@github.com] Sent: November 20, 2015 6:59 PM To: kriswiner/MPU-9250 Subject: [MPU-9250] DMP Usage (#32)
So internal to the MPU-xxxx IMUs, there is this mystical "Digital Motion Processing" unit that does some sensor fusion. I looked through the code and didn't notice it being used. Supposedly it would take out a lot of the pain of doing sensor fusion. Is it due to lack of documentation on the "DMP" unit or why is it you go about getting the raw values then performing your own calculations on them?
Reply to this email directly or view it on GitHub https://github.com/kriswiner/MPU-9250/issues/32 . https://github.com/notifications/beacon/AGY1qih8CJf_v2sYHPD8JSY51E2DxBzQks5 pH9VsgaJpZM4Gm1q3.gif
So internal to the MPU-xxxx IMUs, there is this mystical "Digital Motion Processing" unit that does some sensor fusion. I looked through the code and didn't notice it being used. Supposedly it would take out a lot of the pain of doing sensor fusion. Is it due to lack of documentation on the "DMP" unit or why is it you go about getting the raw values then performing your own calculations on them?