Safe Robot Path Planning and Obstacle Avoidance using Efficient Genetic Algorithm

One of the major drawbacks of the Genetic Algorithm (GA) is the computational complexity due to the random process at each step. A new initial population scheme integrated with a new crossover operator strategy is proposed to overcome this drawback. Before employing the crossover operation, permissible paths based on the c-obstacle concept were generated. To accelerate the convergence, the initial population was divided into two parents, i.e., the parent's chromosome containing the initial and goal positions and the parents composed of nodes from each extracted c-obstacle. Before applying the crossover operator, a filtering algorithm was performed to remove the uncorrelated offspring. A further c-obstacle inclusion made it more efficient; thus, only possible hoping nodes were considered. The random populations and random operations could be reduced efficiently using these steps. Finally, the numerical study method was tested. It is seen that the modified GA is faster and can reduce the total generation, and significantly yields an adaptive generation number.