Synopsis of Soft Computing Techniques used in Quadrotor UAV Modelling and Control

Abstract

The aim of this article is to give an introduction to quadrotor systems with an overview of soft computing techniques used in quadrotor unmanned aerial vehicle (UAV) control, modelling, object following and collision avoidance. The quadrotor system basics, its structure and dynamic model definitions are recapitulated. Further on synopsis is given of previously proposed methods, results evaluated and conclusions drown by authors of referenced publications. The result of this article is a summary of multiple papers on fuzzy logic techniques used in position and altitude control systems for UAVs. Also an overview of fuzzy system based visual servoing for object tracking and collision avoidance is given together with a briefing of quadrotor UAV control techniques efficiency study.Conclusion is that though soft computing methods are widely used with good results, there is still place for much research to be done on find more efficient soft computing tools for simple modelling, robust dynamic control and fast collision avoidance in quadrotor UAV control.

Fuzzy-Genetic Control of Quadrotors Unmanned Aerial Vehicles

“…These six fuzzy-partitions serve as antecedents for the four fuzzy-systems like in equation (11) and (21), used for identifying D ij , with ij = (13,22,23,33) as defined in equations (22)- (24) and (5). Two unknown linear parameters D 11 and D 12 of the quadrotor model as in equation (23), together with 170 linear parameters of the four TSK FLSs (2 FLSs with 5 MFs on one input, each rule with 2 c parameters, plus 2 FLSs with 5 MFs on both of the 2 inputs, each rule with 3c parameters) of equations (22) and equations (24) are determined by the SVD-based LS method as used in [15]. Concluded from equation (17) six fuzzy-partitions (antecedent part of 2 FLSs with 1 input, plus 2 FLSs with 2 inputs are covered by 6 independent fuzzy-partitions) are represented by a vector of six times four K a parameters, which are optimized by a multi-objective hybrid genetic algorithm as detailed in [16].…”

“…Forming fuzzy-partitions by antecedent membership functions ensures that there can not be a numerical input within the defined input range that will not result in firing at least one rule consequent of the fuzzy model, which means that there is a defined output for all possible input states. Keeping specific properties of fuzzy-partitions imposes a set of hard constraints on membership function parameters as detailed in [15]. By imposing these restrictions on all linguistic variables of the FLS, and additionally assuming that the rule base is complete in the sense that it covers the whole input domain, it immediately follows that the TSK model structure of equation (6) simplifies to:…”

“…In [15] a method is introduced that simplifies parameter optimisation of equation (11) while preserving all required constraints. Fuzzy-partitions can be simply formed from Zadeh-typed MFs by making equal the last two parameters of each preceding MF to the first two parameters of the succeeding MF.…”

“…Non-linear parameters can also be simply traced to their local optimum with well-studied gradient descent methods, but we should always keep in mind that gradient descent methods are trapped by local optimum areas. Evolutionary algorithms are robust global optimum search engines, capable of multi-objective search as described in [13][14][15][16].…”

Modeling, control and navigation of an autonomous quad-rotor helicopter

“…This structure can be attractive in several applications, in particular for surveillance, for imaging dangerous environments and for outdoor navigation and mapping [11,12]. The article is organized as follows: Section 1: Introduction.…”