Determining optimum assembly zone for modular reconfigurable robots using multi-objective genetic algorithm

Reconfigurable modular robots can be used in application domains such as exploration, logistics, and outer space. The robots should be able to assemble and work as a single entity to perform a task that requires high throughput. Selecting an optimum assembly position with minimum distance traveled by robots in an obstacle surrounding the environment is challenging. Therefore, this paper proposes a novel approach for optimizing the assembly zone of modular robots in heterogeneous obstacle environments. The method uses a multi-objective Genetic Algorithm (GA) to minimize total travel distance and individual distance disparities. Utilizing the A* algorithm for path planning ensures efficient navigation. A generic kinematic model enabling holonomic locomotion with any reconfiguration and a new modular robot design are also introduced. Hardware experiments have been conducted to validate the kinematic model's applicability for holonomic navigation across different robot configurations. Simulations and physical experiments demonstrated the effectiveness of the proposed method in determining assembly zones, with GA outperforming multi-objective pattern search and random selection in terms of total distance and individual distances traveled by the robots.