🚀 EKF State Estimator

A professional-grade Extended Kalman Filter (EKF) for 2D autonomous vehicle state estimation. This package fuses IMU data (acceleration and yaw rate) with hardware velocity encoders to provide a smooth, continuous, zero-latency odometry estimate.

🏗️ Software Architecture: The Wrapper Pattern

This repository is strictly designed using the Wrapper Pattern to ensure maximum portability, testability, and hardware agnosticism. It is split into two distinct layers:

1. The Pure Math Core (include/EKFCore.hpp)

A pure, vanilla C++ library powered by Eigen3.

• Zero Middleware: Contains absolutely no ROS 2 dependencies.
• Portability: Can be compiled directly into proprietary middleware, embedded systems, or bare-metal microcontrollers tomorrow.
• Math: Implements a dynamic non-holonomic kinematic bicycle model, predicting state via IMU integration and applying measurement updates via high-frequency velocity data.

2. The ROS 2 Wrapper (src/ekf_node.cpp)

A lightweight ROS 2 interface that handles the TF trees, ROS parameter servers, and message serialization.

• Hardware Agnostic: Subscribes to standard geometry_msgs::msg::TwistStamped rather than simulator-specific joint states. This node can be dropped into any robot (drones, rovers, autonomous racecars) that publishes standard ROS velocity topics.
• Configurable: Uses ROS parameters to define dynamic TF frames (odom -> base_link).

---

📡 Node Interfaces

Subscribed Topics

• imu/data (sensor_msgs/Imu): Raw IMU data. The node internally passes this through a Low-Pass Filter (LPF) to eliminate physical hardware vibration and chassis jitter.
• vehicle/velocity (geometry_msgs/TwistStamped): The measured forward velocity from wheel encoders or external sensors.
• ~/reset (std_msgs/Empty): Resets the EKF covariance and state matrices back to origin.

Published Topics

• odometry/filtered (nav_msgs/Odometry): The fused, high-frequency state estimate.
• tf: Broadcasts the transform from odom_frame to base_frame.

ROS 2 Parameters

Parameter	Type	Default	Description
odom_frame	string	"odom"	The parent coordinate frame.
base_frame	string	"base_link"	The child coordinate frame attached to the robot.
publish_tf	bool	true	Whether the node should actively broadcast to /tf.

---

⚙️ Prerequisites

• OS: Ubuntu 22.04
• Middleware: ROS 2 Humble
• Math Library: Eigen3

To install Eigen locally:

sudo apt-get update
sudo apt-get install libeigen-dev