ReVolt State Estimator

A ROS 2 package providing an Error State Extended Kalman Filter (ES-EKF) for estimating a ship’s state (Pose and velocity) by fusing GNSS and IMU data.

NOTE: The current estimate is from NED -> IMU frame since tf2 is not working properly. In theory it is a simple static transformation, but tf2 is struggling to find these transforms so its very slowly. However with the current configuration of the IMU, the imu frame and body frame are the same so no transformation is needed.

tldr; if the imu changes position or orientation, the tf2 transform will need to be implemented, or a hardcoded transform.

---

Overview

The ReVolt State Estimator fuses:

• GNSS (/fix, /heading, /vel)
• IMU (/imu/data)

It outputs:

• TF: NED → Body
• Odometry: nav_msgs/Odometry on topic /state_estimate/revolt

Usage

Nodes will declare parameters from config/revolt_ekf.yaml on startup.

Nodes & Launch

Node: revolt_ekf_node

• Executable: revolt_ekf_node
• Package: revolt_state_estimator
• Launch: revolt_state_estimator.launch.py loads ES-EKF config and starts the node.

Topics & Messages

Subscribed

Topic	Type	Description
/fix	sensor_msgs/NavSatFix	GNSS position (lat, lon + covariance)
/heading	geometry_msgs/QuaternionStamped	GNSS heading (as quaternion)
/vel	geometry_msgs/TwistStamped	GNSS linear velocity (world frame)
/imu/data	sensor_msgs/Imu	IMU orientation, accel & gyro

Published

Topic	Type	Description
/state_estimate/revolt	nav_msgs/Odometry	Estimated [p, v, Theta]
NED → body (TF)	geometry_msgs/TransformStamped	Pose transform

Parameters & Configuration

Configurations live in the config/ folder:

• revolt_ekf.yaml: Noise covariances and topic names

EKF Parameters (revolt_ekf)

Name	Type	Units	Description
Q	[float]	process noise diag ([p, v, b_acc, Theta, b_ars])	Covariance of ins model
R_head	[float]	rad²	GNSS heading noise
R_vel	[float]	[(m/s)², (m/s)², (m/s)²]	GNSS velocity noise

The covariance of the fix is automatically taken from the /fix topic.

State & Measurement Vectors

• State vector x = [p, v, b_acc, Theta, b_ars]:

  • p = [x, y, z] – position in NED (m)
  • v = [v_x, v_y, v_z] – velocity in NED (m/s)
  • b_acc – bias in accelerometer (m/s²)
  • Theta = [phi, theta, psi] – attitude in (rad)
  • b_ars – bias in angular rate sensor (rad/s)

• Measurement:

  • GNSS fix: [x, y, z] position in longitude, latitude, altitude (m)
  • GNSS heading: [yaw] according to north (rad)
  • GNSS vel: [v_x, v_y, v_z] velocity in ENU (m/s)