Revision history [back]

This is not all inclusive answer because the subject it's a very broad and well researched. Excluding hardware demands as this another broad subject, I will just give you some guidance to start your research.

For example ORB_SLAM2 is well established and popular ROS community.

ORB_SLAM2: ROS implementation of the ORB-SLAM2 real-time SLAM library for Monocular, Stereo and RGB-D cameras that computes the camera trajectory and a sparse 3D reconstruction (in the stereo and RGB-D case with true scale). It is able to detect loops and relocalize the camera in real time.

Reference: https://github.com/appliedAI-Initiative/orb_slam_2_ros

To compare the performance against others, you can read ORB-SLAM: a Versatile and Accurate Monocular SLAM System goes in depth about the particulars of benefits, benchmarks and results http://webdiis.unizar.es/~raulmur/MurMontielTardosTRO15.pdf

If you visit KITTI Vision Benchmark Suite: You will find hundreds of implementations, with papers and many algorithms that optimize for particular requirements. http://www.cvlibs.net/datasets/kitti/eval_odometry.php

From this you can use KITTI dataset to test your algorithm and compare performance to others as well. All these implementations can be integrated with ROS.

Other implementation to consider is Open VSLAM: https://arxiv.org/pdf/1910.01122.pdf

And more recent and extremely promising ORB_SLAM3: https://github.com/UZ-SLAMLab/ORB_SLAM3