Factor Graphs and Robust Perception | Michael Kaess | Tartan SLAM Series

A presentation by Michael Kaess as part of the Tartan SLAM Series.
Series overviews and links can be found on our webpage: theairlab.org/tartanslamseries/
Abstract: Factor graphs have become a popular tool for modeling robot perception problems. Not only can they model the bipartite relationship between sensor measurements and variables of interest for inference, but they have also been instrumental in devising novel inference algorithms that exploit the spatial and temporal structure inherent in these problems. I will start with a brief history of these inference algorithms and relevant applications. I will then discuss open challenges in particular related to robustness from the inference perspective and discuss some recent steps towards more robust perception algorithms.
Outline:
0:00 - Welcome & Intro
2:42 - Motivation
4:04 - What are factor graphs?
12:37 - Factor Graphs and Gaussian Inference
19:19 - Smoothing and Mapping
31:45 - Robust Perception: Beyond Gaussian Inference
50:07 - Summary
52:05 - Open Discussion
52:42 - Multimodal sensors in factor graphs?
55:17 - How balance optimization over different factors?
57:19 - How evaluate uncertainty of factors?
1:00:45 - Possible to combine learning and factor graphs?
1:02:25 - Resources to learn about pose graphs?
1:03:17 - How deal with large maps?
1:07:34 - Parallels of SLAM to human learning
RPL Links:
Website: rpl.ri.cmu.edu/
Twitter: / rpl_cmu
AirLab Links:
Website: theairlab.org
Twitter: / airlabcmu
LinkedIn: / the-air-lab-at-carnegi...
Facebook: / airlabcmu
Medium: / airlabcmu

[@antoanbekele9369]

Much appreciated, thank you.

[@urewiofdjsklcmx]

How do you model the uncertainty of the initial position guess of new landmarks in the factor graph? With a prior node?