Control system design optimisation via genetic programming

This paper describes a stochastic approach for comprehensive diagnostics and validation of control system architecture for unmanned aerial vehicle (UAV). Mathematically based diagnostics of a 6 DoF system provides capability for a complex evaluation of system components behaviour, but are typically both memory and computationally expensive. Design and optimisation of the flight controllers is a demanding task which usually requires deep engineering knowledge of intrinsic aircraft behaviour. Evolutionary algorithms (EAs) are known for their robustness for a wide range of optimising functions, when no a priori knowledge of the search space is available. Thus it makes evolutionary approach a promising technique to design the task controllers for complex dynamic systems such as an aircraft. In this study, EAs are used to design a controller for recovery (landing) of a small fixed-wing UAV on a frigate ship deck. The control laws are encoded in a way common for evolutionary programming. However, parameters (numeric coefficients in the control equations) are optimised independently using effective evaluation strategies, while structural changes occur at a slower rate. The fitness evaluation is made via test runs on a comprehensive 6 degree-of-freedom non-linear UAV model. The need of a well defined approach to the control system validation is dictated by the nature of UAV application, where the major source of mission success is based on autonomous control system architecture reliability. The results show that an effective controller can be designed with little knowledge of the aircraft dynamics using appropriate evolutionary techniques. An evolved controller is evaluated and a set of reliable algorithm parameters is validated.