Showing 1–1 of 1 results for author: Kishkun, A

Random Fourier Features based SLAM

Authors: Yermek Kapushev, Anastasia Kishkun, Gonzalo Ferrer, Evgeny Burnaev

Abstract: This work is dedicated to simultaneous continuous-time trajectory estimation and map** based on Gaussian Processes (GP). State-of-the-art GP-based models for Simultaneous Localization and Map** (SLAM) are computationally efficient but can only be used with a restricted class of kernel functions. This paper provides the algorithm based on GP with Random Fourier Features (RFF) approximation for… ▽ More This work is dedicated to simultaneous continuous-time trajectory estimation and map** based on Gaussian Processes (GP). State-of-the-art GP-based models for Simultaneous Localization and Map** (SLAM) are computationally efficient but can only be used with a restricted class of kernel functions. This paper provides the algorithm based on GP with Random Fourier Features (RFF) approximation for SLAM without any constraints. The advantages of RFF for continuous-time SLAM are that we can consider a broader class of kernels and, at the same time, maintain computational complexity at reasonably low level by operating in the Fourier space of features. The accuracy-speed trade-off can be controlled by the number of features. Our experimental results on synthetic and real-world benchmarks demonstrate the cases in which our approach provides better results compared to the current state-of-the-art. △ Less

Submitted 6 September, 2021; v1 submitted 1 November, 2020; originally announced November 2020.