Visual SLAM

Visual simultaneous localization and mapping (vSLAM) is the process of estimating the position and orientation of a camera while simultaneously building a map of its environment using visual inputs. Computer Vision Toolbox™ supports vSLAM workflows for monocular, RGB-D, and stereo cameras, with optional inertial sensor fusion for improved accuracy. These capabilities are essential for applications in robotics, augmented reality, and autonomous navigation. For guidance on choosing a vSLAM workflow, see Choose SLAM Workflow Based on Sensor Data.

Each visual SLAM object—monovslam, rgbdvslam, and stereovslam—provides ready-to-use tools to add frames, track keyframes, compute 3-D map points, estimate camera poses, close loops, and visualize data throughout the camera trajectory. You can also evaluate the performance of the vSLAM algorithm by comparing the estimated camera trajectory to the ground truth using the compareTrajectories function. The toolbox also provides functionality for building your own visual SLAM pipeline.

You can use the toolbox to perform code generation and deployment of vSLAM algorithms. For more information, see Build and Deploy Visual SLAM Algorithm with ROS in MATLAB and Performant and Deployable Monocular Visual SLAM.