SIMULATING AN ARTIFICIAL INTELLIGENCE FOR HELICOPTER FLIGHT MANOEUVRE

Sunny Orike

Department of Electrical/Computer Engineering

Rivers State University of Science and Technology, Port Harcourt-Nigeria

E-mail: orike.sunny@ust.edu.ng,

ABSTRACT

The paper aims to simulate an artificial intelligence for helicopter flight manoeuvre. The methodology makes use of inverse simulation and genetic algorithms to develop a high level helicopter pilot model to fly a prescribed manoeuvre, and realisation of a multi-objective optimisation/search algorithm to converge to a human-like solution. The inverse simulation generates the controls required to fly the helicopter, while the genetic algorithms generate feasible solutions to the inverse simulation problem. The overall goal is to prescribe a manoeuvre for the helicopter and have the developed pilot find control settings that carry out the given manoeuvre. Continuous controls encoding method was implemented in flying an acceleration/deceleration manoeuvre form. The helicopter pilot was formulated as a multi-objective optimization problem with four objectives imposed as penalties. A novel approach, termed maxPenalty, compared and returned the biggest of the four penalties. The genetic algorithm attempts to maximise the fitness function, while minimising the pilot’s total workload. The work evaluates the developed model pilot in terms of performance and functional efficiency.

Keywords: Geneticalgorithms, Helicopter pilot, Inverse Simulation