jcvikl / ekf-slam

EKF-based SLAM implementation using ROS2 & C++. Tested on a real turtlebot 3.

EKF-based SLAM

Implementation of kinematics, odometry, simulation and EKF for SLAM from scratch. Find out more in the linked webpage.

SLAM.video.mov

Package List

This repository consists of several ROS packages

• nuturtle_description - Customized description for the turtlebot.
• nusim - Custom simulator with environment for turtlebot.
• nusim_interfaces - Custom service/message definitions for nusim.
• nuturtle_control - ROS2 control interface for a turtlebot running the custom MSR code.
• nuturtle_control_interfaces - Custom service/message definitions for nuturtle_control.
• nuslam - Feature based EKF SLAM implementation.

There is also a C++ library

• turtlelib - Provides classes and operations for control and sensing.
  • diff_drive - Operations necessary to calculate the kinematics of a differential drive robot.
  • geometry2d - General geometric calculation in 2 dimensions, including vectors, points, angles operations.
  • se2d - Definitions and operations for 2D rigid body motions.
  • svg - Accessory interface to visualize frames.

Odometry test

Odometry.test.-.720p.mov

The odometry test recorded in the video showed a final error of roughly 0.29 meters. This test was performed using encoder odometry solely.

Simulation test

Driving a circle around 4 landmarks in simulation. Green robot represents SLAM estimation, red robot represents ground truth in simulation, blue robot represents odometry-based localization.

2024-03-27.15-04-36.mp4

Real world test

Testing on the actual turtlebot! The SLAM outperforms the odometry with just a couple laps. First video in this readme shows this test.

Driving the robot around the course and taking it back to its initial pose (eye-measuring using a mark on the floor).

Final SLAM estimation (x,y,theta): (-0.02, 0.01, -0.06) Final odometry estimation (x,y,theta): (0.02, 0.24, -0.67)