References & Citations
Computer Science > Robotics
Title: Adaptive Robust Kernels for Non-Linear Least Squares Problems
(Submitted on 30 Apr 2020 (v1), last revised 18 Feb 2021 (this version, v3))
Abstract: State estimation is a key ingredient in most robotic systems. Often, state estimation is performed using some form of least squares minimization. Basically, all error minimization procedures that work on real-world data use robust kernels as the standard way for dealing with outliers in the data. These kernels, however, are often hand-picked, sometimes in different combinations, and their parameters need to be tuned manually for a particular problem. In this paper, we propose the use of a generalized robust kernel family, which is automatically tuned based on the distribution of the residuals and includes the common m-estimators. We tested our adaptive kernel with two popular estimation problems in robotics, namely ICP and bundle adjustment. The experiments presented in this paper suggest that our approach provides higher robustness while avoiding a manual tuning of the kernel parameters.
Submission history
From: Nived Chebrolu [view email][v1] Thu, 30 Apr 2020 16:38:16 GMT (5256kb,D)
[v2] Wed, 27 May 2020 11:06:23 GMT (4938kb,D)
[v3] Thu, 18 Feb 2021 15:13:38 GMT (8171kb,D)
Link back to: arXiv, form interface, contact.