The design of PID controllers for a Gryphon robot using four evolutionary algorithms: a comparative study

• Published in: The Artificial intelligence review Vol. 34; no. 2; pp. 121 - 132
• Main Authors: Vakil-Baghmisheh, Mohammad-Taghi, Salim, Mina
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.08.2010; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Bees; Chromosomes; Comparative studies; Computer Science; Controllers; Design; Evolution; Evolution & development; Evolutionary algorithms; Genes; Genetic algorithms; Methods; Microprocessors; Mutation; Neural networks; Optimization; Particle swarm optimization; Population; Proportional integral derivative; Robot learning; Robots; Step response; Shuffled complex evolution; Nelder–Mead; Gryphon robot; Queen-bee; Particle swarm
• ISSN: 0269-2821; 1573-7462
• DOI: 10.1007/s10462-010-9164-z

This paper compares performances of PID controllers designed for a Gryphon robot joints using four hybrid evolutionary algorithms. We try to reduce the transient state of the step response. To this end, a function of some specifications of the step response (i.e. overshoot, settling time, rise time and the steady state error) is defined. We minimize this function by using four approaches, i.e. Particle swarm optimization (PSO) algorithm, the queen-bee (QB) algorithm, the genetic algorithm (GA) (all hybridized with the Nelder-Mead (NM) algorithm), and the shuffled complex evolution (SCE), and the results are compared. For designing a PID controller we should test several methods and based on the obtained results we can select one of them as the best method. In our case, the queen-bee for joints 1, 2, and 4, the genetic algorithm for joint 3, and the shuffled complex evolution method for joint 5 produce better results.