frequent spikes or jumps in the quaternion data.

[slgrobotics]

Here are my observations while using the "standard" ROS 2 driver on my robot:

1. I have three BNO055 sensors — one from Adafruit with the original Bosch chip, and two “generic carrier boards” using the GY-BNO055 chip.
   Apart from the I²C address (0x28 vs 0x29), there’s no practical difference in how they behave — same data, same issues.

2. The "standard" ROS 2 driver relies on smbus, while a more advanced Python package, smbus2,
   offers improved integration with Linux and better overall reliability.

3. When monitoring /imu/data/orientation/yaw using PlotJuggler, I noticed frequent (~1..3 seconds) spikes or jumps in the data.
   It appears that neither I²C communication nor the code’s exception handling are responsible for this.

Tip: Install PlotJuggler to monitor IMU data:

sudo apt install ros-${ROS_DISTRO}-plotjuggler-ros

Run PlotJugger as a ROS2 process. It subscribes to topic's values (i.e. "imu_data" yaw) and shows them in real time:

ros2 run plotjuggler plotjuggler

I created a fork of the driver, switched to smbus2 and implemented a “sanity check” for incoming data — this fully resolved the issue.

Hopefully the same changes could be merged in the code here, I can create a pull request - let me know.

[slgrobotics]

Opened PR request #83 in case optimizations and examples from my fork are found useful.

[flynneva]

@slgrobotics thanks for the work here. I'll try and review this week and get a new release out 👍🙏

[slgrobotics]

@flynneva great, let me know if I can be of help.