Abstract：For the rapid observation problem in coastal marine environment, an adaptive sampling method based on Gaussian process regression (GPR) for small autonomous underwater vehicle (AUV) is proposed. Firstly, the estimation accuracies and the computational efficiencies are compared among different regression inference methods in GPR, and the sampling interval time is determined. On this basis, GPR analysis is used to predict the environmental data of unobserved areas based on the real-time observation data from AUV, and the AUV is guided to implement online path planning by calculating the regional gradient extremum and the forecasting uncertainty. Finally, this method is used to simulate the regional environmental observation with different feature distributions. Results show that this method can estimate the low-error feature distribution of the observed area more efficiently than the conventional method, can obtain features of the hot spot area more quickly in the observed area, and is more adaptable to the observed area with different feature distributions.
 Kemna S, Caron D A, Sukhatme G S. Adaptive informative sampling with autonomous underwater vehicles:Acoustic versus surface communications[C]//OCEANS 2016. Piscataway, USA:IEEE, 2016:8pp.
 Zeng J B, Li S, Li Y P, et al. Performance of the portable autonomous observation system[C]//OCEANS 2014. Piscataway, USA:IEEE, 2014:4pp.
 Das J, Harvey J, Py F, et al. Hierarchical probabilistic regression for AUV-based adaptive sampling of marine phenomena[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2013:5571-5578.
 Singh A, Krause A, Guestrin C, et al. Efficient planning of informative paths for multiple robots[C]//20th International Joint Conference on Artificial Intelligence. Freiburg, Germany:INST Informatik, 2007:2204-2211.
 Binney J, Krause A, Sukhatme G S. Optimizing waypoints for monitoring spatiotemporal phenomena[J]. International Journal on Robotics Research, 2013, 32(8):873-888.
 Cui R, Li Y, Yan W. Mutual information-based multi-AUV path planning for scalar field sampling using multidimensional RRT*[J]. IEEE Transactions on Systems, Man, and Cybernetics-Systems, 2017, 46(7):993-1004.
 Hitz G, Galceran E, Garneau M, et al. Adaptive continuous-space informative path planning for online environmental monitoring[J]. Journal of Field Robotics, 2017, 34(8):1427-1449.
 Rasmussen C, Williams C. Gaussian processes for machine learning[M]. Cambridge, USA:MIT Press, 2006.
 Ouyang R, Low K H, Chen J, et al. Multi-robot active sensing of non-stationary Gaussian process-based environmental phenomena[C]//International Conference on Autonomous Agents and Multi-agent Systems. Paris, France:International Foundation for Autonomous Agents and Multiagent Systems, 2014:573-580.
 Bai S, Wang J, Chen F, et al. Information-theoretic exploration with Bayesian optimization[C]//IEEE/RSJ International Conference on Intelligent Robots and Systems. Piscataway, USA:IEEE, 2016:1816-1822.
 Ma K C, Liu L, Heidarsson H K, et al. Data-driven learning and planning for environmental sampling[J/OL]. Journal of Field Robotics. (2017-02-07)[2018-03-07]. https://arxiv.org/pdf/1702.01848.pdf. DOI:10.1002/rob.21767.
 Rasmussen C, Nickisch H. Gaussian processes for machine learning (GPML) toolbox[J]. Journal of Machine Learning Research, 2010, 11(6):3011-3015.
 Kemna S, Rogers J G, Nieto-Granda C, et al. Multi-robot coordination through dynamic Voronoi partitioning for informative adaptive sampling in communication-constrained environments[C]//IEEE International Conference on Robotics and Automation. Piscataway, USA:IEEE, 2017:2124-2130.
 曾俊宝,李硕,李一平,等.便携式自主水下机器人控制系统研究与应用[J].机器人,2016,38(1):91-97.Zeng J B, Li S, Li Y P, et al. Research and application of the control system for a portable autonomous underwater vehicle[J]. Robot, 2016, 38(1):91-97.