Multi-Robot Informative Path Planning from Regression with Sparse Gaussian Processes (with Appendix)

TL;DR

An efficient sparse Gaussian process-based approach that uses gradient descent to optimize paths in continuous environments and efficiently scales to both spatially and spatio-temporally correlated environments is proposed.

Abstract

This paper addresses multi-robot informative path planning (IPP) for environmental monitoring. The problem involves determining informative regions in the environment that should be visited by robots to gather the most information about the environment. We propose an efficient sparse Gaussian process-based approach that uses gradient descent to optimize paths in continuous environments. Our approach efficiently scales to both spatially and spatio-temporally correlated environments. Moreover, our approach can simultaneously optimize the informative paths while accounting for routing constraints, such as a distance budget and limits on the robot's velocity and acceleration. Our approach can be used for IPP with both discrete and continuous sensing robots, with point and non-point field-of-view sensing shapes, and for both single and multi-robot IPP. We demonstrate that the proposed approach is fast and accurate on real-world data.

Figures (6)

• Figure 1: An illustration of the expansion and aggregation transformations used in IPP for continuous sensing robots.
• Figure 2: RMSE and runtime results for single robot IPP with the IDP, CIPP, SGP, and Arc-SGP approaches on the ROMS and US soil datasets.
• Figure 3: RMSE and runtime results for four-robot IPP with the CIPP, SGP, and Arc-SGP approaches on the ROMS and US soil datasets.
• Figure 4: Data collection paths generated using a spatio-temporal kernel function for different distance budgets.
• Figure 5: Three different views of our multi-robot IPP solution paths for a spatio-temporal data field, with path lengths of 47.29 m, 47.44 m, and 47.20 m. The total RMSE was 2.75.