# Visual SLAM Using Wheel Odometry

Visual SLAM (Simultaneous Localization and Mapping) traditionally relies on visual features and visual odometry. But in some
mobile robotics use cases—particularly where visual odometry may be unreliable due to motion blur, low texture, or computational
constraints—it is beneficial to fall back to wheel odometry alone. In this guide, we'll explore how to build a Visual SLAM system
without visual odometry, relying entirely on wheel odometry for robot motion estimation. We'll use the Luxonis OAK-D Pro for
stereo vision, iRobot Create3 for mobility and odometry, and an NVIDIA Jetson Xavier NX as the main compute platform—all running
on ROS 2 Humble.

This minimal yet functional setup is ideal for 2D indoor navigation where the environment is flat and wheel slippage is minimal.

## System Configuration Overview

 * Compute NVIDIA Jetson Xavier NX with JetPack 6.0 on Ubuntu 22.04 Tegra
 * Robot Base iRobot Create® 3, flashed with ROS 2 Iron I.0.0
 * Camera Luxonis OAK-D Pro (stereo + IMU)
 * SLAM `rtabmap_ros` with visual odometry disabled
 * Middleware ROS 2 Humble

### Visual SLAM Using Wheel Odometry Blogpost

Read the full write-up on the Visual SLAM Using Wheel odometry workflow.

[Read the blogpost](https://discuss.luxonis.com/blog/6153-visual-slam-using-wheel-odometry-and-luxonis-oak-d-pro-on-ros-2)