My company is upgrading the wheeled odometry system for our driving platform(the old one was inadequate, with massive errors - 10% or so in translation). One of the major components required is the sensor that gives us theta and omega in the XY plane. Our previous module used an ArduIMUV3, which has since been discontinued and also doesn't have excellent performance.

We have also tried out VN100 from VectorNav by borrowing it and that had superior results but isn't really financially feasible. (1500USD!)
We have read good reviews about BNO055, but have also read some opinions that state that it is a bit outdated.

Following are some characteristics of our drive if it can help(Apologies but these are all approximations, not absolute values)

• Maximum Velocity = 6m/s
• Maximum Acceleration = 9m/s2
• Maximum Omega = 360degrees/second
• Vibrations(Z Axis) = >50hz, 2mm amplitude
• Max Runtime = 2-3 Mins per use

Expectations from Sensor

• Minimal Drift during Runtime (< 1 degree)
• Accurate and Precise Omega and Theta Readings (Within 0.1% if possible)
• UART Compatible(It will communicate with an STM32F4, I2C will be second priority)
• High Refresh Rate (Our Control Loop runs at 200hz, so 100hz might work out, will use a pll to get continuous readings if necessary)
• Coprocessor preferable so it has internal filters(to take some load off of STM32)
• Cost Effective (max 40USD)

Thanks a lot in advance!

 

Are you certain on that 10 percent number is down to the module your using.
The mems 9dof sensor market is massive nowadays,
Are you looking for a module with built in power etc, or simple chip solution,
For reference, your current board uses a mpu6000 and a hmc5883l chip.

 

No no, the 10% is due to a lot of different factors which includes mechanical design as well as the algorithms we had on the previous version. I would prefer a chip solution as we are in the process of designing a board for the same, and having the imu integrated in it would be much simpler. (Unless the module has some other functional advantages that I'm not aware of)

 

No no, the 10% is due to a lot of different factors which includes mechanical design as well as the algorithms we had on the previous version. I would prefer a chip solution as we are in the process of designing a board for the same, and having the imu integrated in it would be much simpler. (Unless the module has some other functional advantages that I'm not aware of)

Sensor accuracy is achieved in more modern sensors by sensor fusion,
You will find d that a module will have they cide to achieve this its 3d maths and eulers , all a bit fun maths.
So the manatomiter , inertial sensors , gps etc all are fused to give a more accurate reading.

 

I think a module would be a much better option then, can you recommend any that can suit our applications?

 

have look here
https://community.bosch-sensortec.c...est-performance/td-p/25233?attachment-id=2958
what worries me, is that the sensor you have should easily be able to get much better then you are achieving,
and unless this addressed, your not going to get an improvment,
Id strongly suggets that you look at the arduino modues, an arduino and module is around 40 dollars, you can afford to get a few different ones and experiment, and see which suits your needs,
Id say any module will meet your needs,

 

I agree, maybe the issue is with our own implementation of the module. I'll take a look at it. We are currently designing our custom odometry board with stm so will have to integrate with that. The site you have mentioned talks about the BMI series, what is the difference between that and the BNO series?

 

I agree, maybe the issue is with our own implementation of the module. I'll take a look at it. We are currently designing our custom odometry board with stm so will have to integrate with that. The site you have mentioned talks about the BMI series, what is the difference between that and the BNO series?

sorry dont know