IMU

Hardware

AutomatePro includes an onboard BNO085 9-axis MEMS IMU. The IMU combines a 3-axis accelerometer, 3-axis gyroscope, and 3-axis magnetometer. The driver publishes ROS 2 IMU data and, when enabled, magnetic-field and accuracy data.

The arrows on the device indicate the IMU vector directions:

IMU Specifications

Parameter	Value
Sensor	BNO085
Sensing types	3-axis gyroscope, 3-axis magnetometer, 3-axis accelerometer
Rotation vector error (dynamic)1	3.5 deg
Accelerometer accuracy1	0.3 m/s2
Gyroscope accuracy1	3.1 deg/s
Magnetometer accuracy1	1.4 uT
Default IMU sample rate	100 Hz
Configurable IMU rate	Up to 400 Hz, hardware dependent

Info

¹ Based on IC manufacturer tests. Results on AutomatePro can vary by mounting, calibration, and magnetic environment. See the BNO080/BNO085 datasheet for manufacturer details.

ROS API

Node: automatepro_imu_driver

Publishers

Topic	Type	Description
/sensor/imu/data	sensor_msgs/msg/Imu	Orientation, angular velocity, and linear acceleration.
/sensor/imu/magnetic_field	sensor_msgs/msg/MagneticField	Magnetic field data, when magnetometer publishing is enabled.
/sensor/imu/accuracy	automatepro_imu_driver/msg/ImuAccuracy	Accuracy levels for orientation, accelerometer, gyroscope, and magnetic-field reports.

Services

Service	Type	Description
/sensor/imu/save_tare	std_srvs/srv/Trigger	Persists the current tare to sensor flash.
/sensor/imu/save_calibration	std_srvs/srv/Trigger	Saves Dynamic Calibration Data (DCD) to sensor flash.
/sensor/imu/reinit	std_srvs/srv/Trigger	Reinitializes the IMU driver and sensor.

Verify IMU Data

Run these checks from AutomatePro over SSH:

source /opt/ros/humble/setup.bash
ros2 node list | grep automatepro_imu_driver
ros2 topic echo /sensor/imu/data --once --qos-reliability best_effort

If accuracy publishing is enabled, check calibration state:

ros2 topic echo /sensor/imu/accuracy --once

Accuracy levels use:

Value	Meaning
0	Unreliable
1	Low
2	Medium
3	High
255	Unknown

Configuration

Driver: automatepro_imu_driver
Container: automatepro-core-driver

info

Runtime ROS configuration is stored in /opt/automatepro/config/ros. Packaged defaults are stored in /opt/automatepro/defconfig/ros.

Edit the IMU parameter file:

sudo nano /opt/automatepro/config/ros/imu_params.yaml

Important parameters:

Parameter	Default	Description
log_level	info	Driver log level. Launch accepts debug, info, warn, error, or fatal.
frame_id	lm450_imu_link	Frame ID stamped on published IMU messages.
rotation_vector_mode	game_rotation_vector	Orientation report type. Use rotation_vector for magnetic absolute heading, game_rotation_vector for relative heading without magnetometer, or geomagnetic_rotation_vector for geomagnetic heading.
tare.on_startup	false	Applies one-shot tare when the node starts. Leave disabled after commissioning.
tare.axes	xyz	Axes included in tare.
tare.persist	false	Saves tare to flash when enabled.
tare.use_persisted	false	Composes the tare saved in the sensor with the mounting orientation. When false, the driver ignores the saved tare and uses the mounting orientation as the absolute reorientation.
tare.orientation.enabled	true	Applies the configured chip-to-body orientation mapping at startup and watchdog reinitialization.
tare.orientation.roll_deg/pitch_deg/yaw_deg	0.0/0.0/0.0	Mounting orientation in degrees when use_quaternion is false.
tare.orientation.use_quaternion	false	Uses qx, qy, qz, and qw instead of roll, pitch, and yaw when enabled.
tare.orientation.qx/qy/qz/qw	0/0/0/1	Quaternion orientation mapping when use_quaternion is true.
calibration.accelerometer.enabled	false	Enables BNO08x accelerometer dynamic calibration.
calibration.accelerometer.planar	false	Enables planar accelerometer calibration mode.
calibration.gyroscope.enabled	false	Enables BNO08x gyroscope dynamic calibration.
calibration.gyroscope.on_table	false	Enables stationary on-table gyroscope calibration assist.
calibration.magnetic_field.enabled	false	Enables BNO08x magnetometer dynamic calibration. Enable only when magnetic heading is required and the environment is suitable.
calibration.save_dcd	false	Periodically saves Dynamic Calibration Data to flash.
i2c.enabled	true	Uses I2C communication.
i2c.bus	/dev/i2c-7	I2C bus path.
i2c.address	0x4B	BNO08x I2C address.
uart.enabled	false	Uses UART communication instead of I2C when enabled and i2c.enabled is false.
uart.device	/dev/ttyACM0	UART device path when UART transport is enabled.
publish.accuracy.enabled	true	Publishes /sensor/imu/accuracy.
publish.imu.enabled	true	Publishes /sensor/imu/data.
publish.imu.rate	100	Requested IMU report rate in Hz.
publish.magnetic_field.enabled	false	Publishes /sensor/imu/magnetic_field. Use --qos-reliability best_effort when echoing this topic from the CLI.
publish.magnetic_field.rate	100	Requested magnetometer report rate in Hz.
publish.imu.orientation_offset.enabled	false	Applies a software quaternion offset before publishing orientation.
publish.imu.orientation_offset.qx/qy/qz/qw	0/0/0/1	Software orientation offset quaternion.

Important

Restart automatepro-core-driver after changing the configuration. Configuration values are loaded when the driver container starts and are only applied after the service restarts.

sudo systemctl restart automatepro-core-driver

Default Configuration Shape

automatepro_imu_driver:
  ros__parameters:
    log_level: "info"
    frame_id: "lm450_imu_link"
    rotation_vector_mode: "game_rotation_vector"
    tare:
      on_startup: false
      axes: "xyz"
      persist: false
      use_persisted: false
      orientation:
        enabled: true
        roll_deg: 0.0
        pitch_deg: 0.0
        yaw_deg: 0.0
        use_quaternion: false
        qx: 0.0
        qy: 0.0
        qz: 0.0
        qw: 1.0
    calibration:
      accelerometer:
        enabled: false
        planar: false
      gyroscope:
        enabled: false
        on_table: false
      magnetic_field:
        enabled: false
      save_dcd: false
    i2c:
      enabled: true
      bus: "/dev/i2c-7"
      address: "0x4B"
    uart:
      enabled: false
      device: "/dev/ttyACM0"
    publish:
      accuracy:
        enabled: true
      magnetic_field:
        enabled: false
        rate: 100
      imu:
        enabled: true
        rate: 100
        orientation_offset:
          enabled: false
          qx: 0.0
          qy: 0.0
          qz: 0.0
          qw: 1.0

Example

This example prints IMU and magnetic field messages, each at most once per second, by subscribing to /sensor/imu/data and /sensor/imu/magnetic_field; the driver publishes the magnetic field only when publish.magnetic_field.enabled is true. The tutorial package is available in the AutomatePro tutorials repository, and its README explains how to build and run the examples.

Python

Source: imu.py

import rclpy
from rclpy.executors import ExternalShutdownException
from rclpy.node import Node
from sensor_msgs.msg import Imu, MagneticField

LOG_PERIOD_S = 1.0


class ImuSubscriber(Node):

    def __init__(self):
        super().__init__('imu_subscriber')
        self.imu_subscription = self.create_subscription(
            Imu,
            '/sensor/imu/data',
            self.imu_callback,
            10)
        self.magnetic_field_subscription = self.create_subscription(
            MagneticField,
            '/sensor/imu/magnetic_field',
            self.magnetic_field_callback,
            10)

    def imu_callback(self, msg):
        self.get_logger().info(
            'Received IMU message: orientation=%s, angular_velocity=%s, '
            'linear_acceleration=%s' % (
                msg.orientation, msg.angular_velocity, msg.linear_acceleration),
            throttle_duration_sec=LOG_PERIOD_S)

    def magnetic_field_callback(self, msg):
        self.get_logger().info(
            'Received MagneticField message: magnetic_field=%s' % msg.magnetic_field,
            throttle_duration_sec=LOG_PERIOD_S)


def main(args=None):
    rclpy.init(args=args)
    node = ImuSubscriber()
    try:
        rclpy.spin(node)
    except (KeyboardInterrupt, ExternalShutdownException):
        pass
    finally:
        node.destroy_node()
        rclpy.try_shutdown()


if __name__ == '__main__':
    main()

Run the node using the following command:

ros2 run automatepro_python_tutorials imu_node

C++

Source: imu.cpp

#include <rclcpp/rclcpp.hpp>
#include <sensor_msgs/msg/imu.hpp>
#include <sensor_msgs/msg/magnetic_field.hpp>

class ImuSubscriber : public rclcpp::Node
{
public:
  ImuSubscriber()
  : Node("imu_subscriber")
  {
    imu_subscription_ = this->create_subscription<sensor_msgs::msg::Imu>(
      "/sensor/imu/data", 10, std::bind(&ImuSubscriber::imu_callback, this, std::placeholders::_1));

    magnetic_field_subscription_ = this->create_subscription<sensor_msgs::msg::MagneticField>(
      "/sensor/imu/magnetic_field", 10,
      std::bind(&ImuSubscriber::magnetic_field_callback, this, std::placeholders::_1));
  }

private:
  static constexpr int kLogPeriodMs = 1000;

  void imu_callback(const sensor_msgs::msg::Imu::SharedPtr msg)
  {
    RCLCPP_INFO_THROTTLE(
      this->get_logger(),
      *this->get_clock(), kLogPeriodMs,
      "Received IMU message: orientation=[x: %f, y: %f, z: %f, w: %f], "
      "angular_velocity=[x: %f, y: %f, z: %f], linear_acceleration=[x: %f, y: %f, z: %f]",
      msg->orientation.x, msg->orientation.y, msg->orientation.z, msg->orientation.w,
      msg->angular_velocity.x, msg->angular_velocity.y, msg->angular_velocity.z,
      msg->linear_acceleration.x, msg->linear_acceleration.y, msg->linear_acceleration.z);
  }

  void magnetic_field_callback(const sensor_msgs::msg::MagneticField::SharedPtr msg)
  {
    RCLCPP_INFO_THROTTLE(
      this->get_logger(), *this->get_clock(), kLogPeriodMs,
      "Received MagneticField message: magnetic_field=[x: %.3e, y: %.3e, z: %.3e]",
      msg->magnetic_field.x, msg->magnetic_field.y, msg->magnetic_field.z);
  }

  rclcpp::Subscription<sensor_msgs::msg::Imu>::SharedPtr imu_subscription_;
  rclcpp::Subscription<sensor_msgs::msg::MagneticField>::SharedPtr magnetic_field_subscription_;
};

int main(int argc, char * argv[])
{
  rclcpp::init(argc, argv);
  rclcpp::spin(std::make_shared<ImuSubscriber>());
  rclcpp::shutdown();
  return 0;
}

Run the node using the following command:

ros2 run automatepro_cpp_tutorials imu_node

Orientation Commissioning

Use this workflow when the IMU is remounted, when the reference orientation changes, or when a system needs a known rest pose.

Important

Perform tare and calibration only when the device is in its final mounting position and away from motors, steel structures, strong magnets, and current-carrying wires.

1. Complete dynamic calibration.

   Dynamic calibration is off in the packaged configuration, because slow yaw turns during operation mis-calibrate the gyroscope. For a calibration session, set calibration.accelerometer.enabled and calibration.gyroscope.enabled to true, and calibration.magnetic_field.enabled only if magnetic heading is required, then restart the core driver.

   Move the device through several stable orientations for accelerometer calibration, leave it stationary for gyroscope calibration, and perform magnetic rotations only if magnetic-field calibration is enabled. Save the calibration:

   ros2 service call /sensor/imu/save_calibration std_srvs/srv/Trigger

   Set the calibration flags back to false and restart the core driver.

2. Tare the reference pose (optional).

   With the packaged tare.use_persisted: false, the driver ignores a tare saved in the sensor and uses the mounting orientation from step 3 as the absolute reorientation, so a tare does not survive a restart. A saved tare does not apply to game_rotation_vector, the default mode; use steps 3 and 4 instead.

   To use a saved tare with rotation_vector, place AutomatePro in the desired zero orientation and temporarily set:

   tare:
     on_startup: true
     axes: "xyz"
     persist: true
     use_persisted: true

   Restart the core driver and wait for tare completion in logs:

   sudo systemctl restart automatepro-core-driver
   sudo journalctl -u automatepro-core-driver -f

   After tare completes, set tare.on_startup: false so it is not applied on every boot. Keep tare.use_persisted: true.

3. Set mounting orientation.

   Use tare.orientation.roll_deg, pitch_deg, and yaw_deg to map the chip axes to the device body frame. This orientation mapping is re-applied on startup and watchdog recovery.

4. Apply software heading offset when needed.

   If the device in the rest pose should publish identity orientation (0, 0, 0, 1), capture the current quaternion:

   ros2 topic echo /sensor/imu/data --field orientation --once

   Set publish.imu.orientation_offset.* to the conjugate of that quaternion and enable publish.imu.orientation_offset.enabled.

5. Verify the final output.

   Restart the core driver and verify that the rest pose and motion directions match your robot frame:

   sudo systemctl restart automatepro-core-driver
   ros2 topic echo /sensor/imu/data --field orientation --once

Developer Notes: rotation modes and persistence

rotation_vector_mode selects which BNO08x SH-2 report is enabled. game_rotation_vector does not use the magnetometer and reports relative heading. rotation_vector and geomagnetic_rotation_vector use magnetic information for heading.

Dynamic tare runs only on initial node startup when tare.on_startup is enabled. The fixed mounting orientation is re-applied on every sensor initialization, including watchdog recovery. The software orientation offset is applied as q_published = q_offset * q_sensor.

Troubleshooting

IMU Topic Is Missing

Check that the driver is enabled:

grep ROS_DRIVER_IMU /opt/automatepro/.env

Check logs:

sudo journalctl -u automatepro-core-driver -n 100 --no-pager

Automatic Recovery

The driver watches the sensor data stream and reinitializes the sensor when no data arrives for two seconds. This is the same action as the /sensor/imu/reinit service, and the period and timeout are fixed in the driver.

The mounting orientation configured under tare.orientation is applied again on every reinitialization. A dynamic tare is not, as described in Orientation Commissioning.

Orientation Is Correct After Startup but Wrong After Recovery

Verify tare.orientation.enabled: true. The fixed orientation mapping is the part reapplied on every sensor initialization.

Heading Drifts or Jumps

Check the rotation-vector mode and magnetic environment. If the system does not require magnetic absolute heading, game_rotation_vector is less sensitive to magnetic disturbances. If magnetic heading is required, calibrate in the deployment environment and monitor /sensor/imu/accuracy.