Optimal path planning of multiple nanoparticles in continuous environment using a novel Adaptive Genetic Algorithm

Abstract This paper presents a novel Adaptive Genetic Algorithm for optimal path planning of multiple nanoparticles during the nanomanipulation process. The proposed approach determines the optimal manipulation path in the presence of surface roughness and environment obstacles by considering constraints imposed on the nanomanipulation process. In this research, first by discretizing the environment, an initial set of feasible paths were generated, and then, path optimization was continued in the original continuous environment (and not in the discrete environment). The presented novel approach for path planning in continuous environment (1) makes the algorithm independent of grid size, which is the main limitation in conventional path planning methods, and (2) creates a curve path, instead of piecewise linear one, which increases the accuracy and smoothness of the path considerably. Every path is evaluated based on three factors: the displacement effort (the area under critical force-time diagram during nanomanipulation), surface roughness along the path, and smoothness of the path. Using the weighted linear sum of the mentioned three factors as the objective function provides the opportunity to (1) find a path with optimal value for all factors, (2) increase/decrease the effect of a factor based on process considerations. While the former can be obtained by a simple weight tuning procedure introduced in this paper, the latter can be obtained by increasing/decreasing the weight value associated with a factor. In the case of multiple nanoparticles, a co-evolutionary adaptive algorithm is introduced to find the best destination for each nanoparticle, the best sequence of movement, and optimal path for each nanoparticle. By introducing two new operators, it was shown that the performance of the presented co-evolutionary mechanism outperforms the similar previous works. Finally, the proposed approach was also developed based on a modified Particle Swarm Optimization algorithm, and its performance was compared with the proposed Adaptive Genetic Algorithm.