Random Fourier Features based SLAM

This work is dedicated to simultaneous continuous-time trajectory estimation and mapping based on Gaussian Processes (GP). State-of-the-art GP-based models for Simultaneous Localization and Mapping (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.

Authors

• Yermek Kapushev
• Anastasia Kishkun
• Gonzalo Ferrer 0001
• Evgeny Burnaev

Keywords

• Weight measurement
• Simultaneous localization and mapping
• Gaussian processes
• Benchmark testing
• Trajectory
• Sparse matrices
• Noise measurement
• Random Feature
• Fourier Features
• Random Fourier Features
• Computational Complexity
• Computational Efficiency
• Functional Class
• Kernel Function
• Gaussian Process
• Simultaneous Mapping
• Trajectory Estimation
• Trajectory Mapping
• Covariance Matrix
• Feature Maps
• Invertible
• Radial Basis Function
• Kriging
• Radial Basis Function Kernel
• Maximum A Posteriori
• Landmark Localization
• Number Of Landmarks
• Translation Error
• Gaussian Process Model
• Ground Truth Trajectory
• Rotation Error
• Matrix-vector Product
• Camera Pose
• Gaussian Process Regression Model
• Stereo Images
• Iterative Solver