SCIENCE CHINA Information Sciences, Volume 62 , Issue 5 : 050208(2019) https://doi.org/10.1007/s11432-018-9653-7

Long-term adaptive informative path planning for scalar field monitoring using cross-entropy optimization

More info
  • ReceivedSep 14, 2018
  • AcceptedOct 7, 2018
  • PublishedFeb 27, 2019


There is no abstract available for this article.


This work was supported by National Natural Science Foundation of China (Grant Nos. U1813225, 61472325, 61733014, 51579210) and Science, Technology and Innovation Commission of Shenzhen Municipality (Grant No. JCYJ20170817145216803).


[1] Li Z, Huang B, Ye Z. Physical Human-Robot Interaction of a Robotic Exoskeleton By Admittance Control. IEEE Trans Ind Electron, 2018, 65: 9614-9624 CrossRef Google Scholar

[2] Pan Y, Yang C, Pan L. Integral Sliding Mode Control: Performance, Modification, and Improvement. IEEE Trans Ind Inf, 2018, 14: 3087-3096 CrossRef Google Scholar

[3] Ma K C, Liu L, Heidarsson H K. Data-driven learning and planning for environmental sampling. J Field Robotics, 2018, 35: 643-661 CrossRef Google Scholar

[4] Yang C, Jiang Y, He W. Adaptive Parameter Estimation and Control Design for Robot Manipulators With Finite-Time Convergence. IEEE Trans Ind Electron, 2018, 65: 8112-8123 CrossRef Google Scholar

[5] Seeger M. Gaussian processes for machine learning.. Int J Neur Syst, 2004, 14: 69-106 CrossRef PubMed Google Scholar

[6] Yang K, Moon S, Yoo S. Spline-Based RRT Path Planner for Non-Holonomic Robots. J Intell Robot Syst, 2014, 73: 763-782 CrossRef Google Scholar

[7] Reuven R. The cross-entropy method for combinatorial and continuous optimization. Methodol Comput Appl Probabil, 1999, 1: 127--190. Google Scholar


    Algorithm 1 Long-term adaptive IPP algorithm

    Require:starting point $x_{\rm~start}$, planning horizon $\varrho$, historical sequence of control points $X_H$, obstacle region $\mathcal{X}_{\rm~obs}$.

    Output:posterior estimation of scalar field $\tilde{f}$, local optimal path $\tau_{X_C}$.


    local optimal sequence of control points distribution $[\mu^*,s_2^*]\leftarrow~{\rm~CEoptimize}(\mu,s_2,\mathcal{X}_{\rm~obs},c_{\rm~left},X_{T})$;

    $X_C\leftarrow~{\rm~Sample}(\mu^*,s_2^*)$; $X_F\leftarrow\tau_{X_C}$;


    adaptive re-plan $X_{T_d}\leftarrow~{\rm~Sample}(p(x_{T_d}))$.


    Algorithm 2 Cross-entropy optimization algorithm

    and sort as $O_1\leq~O_2~\leq\cdots\leq~O_N$; $\gamma_t\leftarrow~O_{\lceil(1-\eta)N\rceil}$;


    Require:$v_1$, quantile $\eta$, size $N$, max iterations $M$.

    Output:pdf of $x^*$, i.e., $g(v^*)$.

    for $t=2,\ldots,M$

    generate samples from pdf $g(X,v_{t-1})$,