Reconstruction of radiation dose rate profiles by autonomous robot with active learning and Gaussian process regression

Abstract This work proposes an approach to use autonomous intelligent robots to reconstruct 2D dose rate profiles in radioactive environments. The main idea is to use Gaussian process (GP) regression to model the radiation dose rate distribution and active learning (AL) to dynamically improve estimation of GP parameters. A differential evolution (DE) algorithm, guided by the GP entropy is used to find the next measuring point to be visited by the robot in order to improve the GP estimation. Euclidean distances are used as penalizations to minimize the robot’s trajectory and, consequently, to speed-up the reconstruction of the dose rate profile. The intelligent algorithm was tested through computational simulations considering 2 hypothetical scenarios with increasing complexity. Results demonstrate that the approach is able to reconstruct the dose rate profile with good accuracy and reduced number of measurements.

[1]  Eijiro Takeuchi,et al.  Robotic control vehicle for measuring radiation in Fukushima Daiichi Nuclear Power Plant , 2011, 2011 IEEE International Symposium on Safety, Security, and Rescue Robotics.

[2]  Carl E. Rasmussen,et al.  Gaussian Processes for Data-Efficient Learning in Robotics and Control , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Rainer Storn,et al.  Differential Evolution – A Simple and Efficient Heuristic for global Optimization over Continuous Spaces , 1997, J. Glob. Optim..

[4]  Dongbing Gu,et al.  Active learning of Gaussian processes for spatial functions in mobile sensor networks , 2011 .

[5]  Nando de Freitas,et al.  A Bayesian exploration-exploitation approach for optimal online sensing and planning with a visually guided mobile robot , 2009, Auton. Robots.

[6]  Andrew Graham,et al.  Snaking around in a nuclear jungle , 2005, Ind. Robot.

[7]  Bing Lam Luk,et al.  Intelligent legged climbing service robot for remote maintenance applications in hazardous environments , 2005, Robotics Auton. Syst..

[8]  Carl E. Rasmussen,et al.  Warped Gaussian Processes , 2003, NIPS.

[9]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[10]  Agathoniki Trigoni,et al.  An Underwater Robotic Network for Monitoring Nuclear Waste Storage Pools , 2009, S-CUBE.

[11]  Cyrill Stachniss,et al.  Learning gas distribution models using sparse Gaussian process mixtures , 2009, Auton. Robots.

[12]  Miguel A. Serna,et al.  Wall-climbing robot for inspection in nuclear power plants , 1994, Proceedings of the 1994 IEEE International Conference on Robotics and Automation.