@phdthesis{oai:kyutech.repo.nii.ac.jp:00006273,
 author = {Maroofkhani,  Farhad},
 month = {2019-12-10},
 note = {1 Introduction||2 Levy Walk||3 Design of Experiment （DOE）||4 Response Surface Methodology||5 Results and Discussions||6 Conclusion, Autonomous robot’s search strategy is the set of rules that it employs while looking for targets in its environment. In this study, the stochastic movement of robots in unknown environments is statistically studied, using a Levy walk method. Biological systems （e.g., foraging animals） provide useful models for designing optimal stochastic search algorithms. Observations of biological systems, ranging from large animals to immune cells, have inspired the design of efficient search strategies that incorporate stochastic movement. In this study, we seek to identify the optimal stochastic strategies for autonomous robots. Given the complexity of interaction between the robot and its environment, optimization must be performed in high-dimensional parameter space. The effect of the explanatory variable on the forger robot movement with the minimum required energy was also studied using experiments done by the response surface methodology （RSM）. We analyzed the extent to which search efficiency requires these characteristics, using RSM. Correlation between the involved parameters via a Lévy walk process was examined through designing a setup for the experiments to determine the interaction of the involved variables and the robot movement. The extracted statistical model represents the priority influence of those variables on the robot by developing the statistical model of the mentioned unknown area. The efficiency of a simple strategy was investigated based on Lévy walk search in two-dimensional landscapes with clumped resource distributions. We show how RSM techniques can be used to identify optimal parameter values as well as to describe how sensitive efficiency reacts to the changes in these values. Here, we identified optimal parameter for designing robot by using stochastic search pattern and applying mood-switching criteria on a mixture of speed and sensor and μ to determine how many robots are needed for a solution. Fractal criterion-based robot strategies were more efficient than those based on the resource encounter criterion, and the former was found to be more robust to changes in resource distribution as well., 九州工業大学博士学位論文 学位記番号:生工博甲第358号 学位授与年月日:令和元年9月20日, 令和元年度},
 school = {九州工業大学},
 title = {Design and Optimization of Stochastic Search Strategy for Foraging Problem},
 year = {}
}