Robotic Application of Particle Swarm Optimization Algorithm

Swarm algorithms are generally used to find the optimal solution to a given problem using multiple particles that collect data. This swarm algorithm is executed by multiple robots that search for the hottest location on an arena by communicating and working together. These robots use their current direction, the location of their hottest recorded temperature, the location of the swarm’s hottest recorded temperature, and a random vector to determine where to search next. By varying these weighted components, this application seeks an optimal algorithm for locating the hottest spot on the arena in an efficient manner. This kind of work has thus far been limited to study in the realm of computer science. Our work is some of the first to expand this research to actual robotics. We expect to learn how this algorithm must be modified to be practical in real life implementations. A few complications that we are particularly interested in include the implication of real space as opposed to the points or pixels to which theoretical algorithms are limited. The sensitivity of the robots to hot spots, both those found by the swarm and those found by the individual robots, will determine the efficiency of the algorithm, and by varying these sensitivities we can find an optimal algorithm. Lastly, we hope to optimize the algorithm by stratifying the temperatures of hot spots and distributing jobs among the robots such that each robot is exploring a different strata of temperature