Swarm object transportation through phase transitions

• Main Author: Chua, Joseph Aldrin T.
• Format: text
• Language: English
• Published: Animo Repository 2021
• Subjects: Swarm intelligence; Robots—Control systems; Mechanical Engineering
• Online Access: https://animorepository.dlsu.edu.ph/etdm_mecheng/3
• Institution: De La Salle University

Controlling Multiple Robot Systems (MRS) has been the hindrance towards the favoring of MRS over single robot systems in different application fields. Despite the possible advantages of MRS, controlling each agent in the system towards a common objective remains to be a challenge. Swarm robotic systems, in particular, has long been based on biological swarms and more recently in physics concepts to control each agent. The use of phase transitions allows the swarm to limit to operating with only three different control modes, depending on the objective of the system. Cooperative object transportation is a common task for robot systems. This study focuses on two aspects of cooperative object transportation, moving toward the object and moving the object itself. The intended behavior of the swarm for this task is that it will initially take a liquid phase to move through obstacles efficiently then transitioning to a solid to cooperatively transport the object. The study tests both aspects separately. In the liquid phase, the objective of the swarm is to avoid collisions with obstacles and other members using the Moving Particle Semi-Implicit (MPS) method. The transportation of the object focuses on applying a proposed transportation method for the swarm testing the effect of different formations in the movement of the swarm. The MPS method is a particle method used to simulate fluid flow. Developed in 1996 by Koshizuka and Oka, the method is used for incompressible fluids. In swarm robots, another particle method has already been used in other literatures called the Smoothed Particle Hydrodynamics (SPH) technique. This method, however, was initially developed for compressible fluids. As the desired behavior is to have the swarm elicit the liquid phase, the MPS method can be a better fit for the intended application. Particles and robots, however, require different constraint considerations. For instance, particles tend to keep accelerating while robots are constrained by its motor speed. Such considerations were applied in the study, modifying the original MPS method for swarm control. Results of the swarm in the liquid phase showed few instances of collisions that can be remedied by adjusting parameters such as collision distance and particle size. In the solid phase, different formations showed different trajectories for an intended straight-line path. Given the current setup, a different object attachment method must be used for a better performance in object transportation.