Alignment and scale drift is usually solved via loop closure. Clearer example in the LSD-SLAM: https://www.youtube.com/watch?v=GnuQzP3gty4&feature=youtu.be&t=80

Alignment and scale drift is usually solved via loop closure. Clearer example in the LSD-SLAM: https://www.youtube.com/watch?v=GnuQzP3gty4&feature=youtu.be&t=80 https://www.youtube.com/watch?v=GnuQzP3gty4&feature=youtu.be.... By the way it seems like the researcher behind DSO and LSD-SLAM is now at oculus research! http://vision.in.tum.de/members/engelj http://vision.in.tum.de/members/engelj