Why can't you "process" the incoming data from the MPU9150 via a script to smooth and filter out the "noise"? I would assume the data from the iPhone's sensors is already pre-processed that's why it appears more stable....

One thread is/was good enough... no need to cross post we heard you.....

Dave, what numbers are all over the place? How can I help you get up and running?

The compass headings from the MPU just seem to be very erratic compared to the sensorstream. I am going to be using the EZB-v4 in an essentially fully autonomus bot reacting to two moving objects in an outdoor environment. As one object or the other approach the bot I need it move to a fairly accurate heading.

As an aside will a longer range ultrasonic rangefinder such as the maxbotic line work pretty much plug and play?

Also has anyone done any interfacing with a Lidar unit such as the hokuyo 30UTM-lX? I can import the data into ARC from Roborealm which has a dedicated module for the scanner but would a direct feed into ARC be a better method

Sorry for the double post Richard R...very nice way to introduce yourself to a total neophyte to robotics.

Thanks for the help, Dave

@Drdaveyo What you can do is take a handful of readings from the MPU9150 and avg them .... e.g... maybe take five readings per second and then divide by 5.... this should give you a fairly accurate direction reading every second or so...

I have attached a couple videos of the output when the board is rotated....one is the sensor stream sending about 60 packets /sec. The other is the MPU....

a 1 sec update won't be nearly fast enough. If I know I have to use the sensorstream that is fine....I just figured the board ( or SOME board) would have comparable performance...maybe it is not working properly.

  Sensor stream   

MPU

You can poll much faster...1sec was just my example.... BTW I am willing to bet the sensor stream from your iPhone has already been processed.... I doubt it is raw data from the on board sensor(s).... Just saying....