Decentralized Sparse Gaussian Process Regression with Event-Triggered Adaptive Inducing Points

In this paper, we present a decentralized sparse Gaussian process regression (DSGPR) model with event-triggered, adaptive inducing points. We address common issues inherent in decentralized frameworks such as high computation costs, inter-agent message bandwidth restrictions, and data fusion integrity. We also improve real-time sparse Gaussian process regression models by introducing an adaptation of the mean shift and fixed-width clustering algorithms called radial clustering. Radial clustering enables real-time SGPR models to have an adaptive number of inducing points through an efficient inducing point selection process. The entire DSGPR framework is evaluated on multiple simulated random vector fields. The results show that this framework effectively estimates vector fields using multiple autonomous agents. We also show the practical use of these algorithms by testing them on hardware typical for small autonomous vehicles