Domain knowledge based genetic algorithms for mobile robot path planning having single and multiple targets

Abstract

In path planning problem of a mobile robot, the objective is to find an optimal collision free path from a source to single or multiple targets. Domain knowledge based genetic algorithms have been proposed to address the path planning problem having single as well as multiple independent targets. Four new domain knowledge based operators namely “circuit removal operator”, “insertion-deletion operator”, “refinement Operator”, and “target alignment operator” have been introduced in this work. Among these four operators, first three have been used for the path planning problem having single target, whereas all these four operators have been used for path planning problem having multiple independent targets. Proposed methods have been deployed on several simulated environments of varying sizes. From the experimental results, it has been established that the domain knowledge based operators enhance the capability of conventional genetic algorithm. Our proposed method for Mobile Robot’s Path Planning Problem having single target has outperformed some previously proposed evolutionary algorithm based methods.

Keywords

• Genetic algorithms
• Robot path planning
• Domain knowledge based operator
• Single target path planning
• Multiple targets path planning