Article Preview
Top1. Introduction
Unmanned Vehicles (UAVs) are the most suitable platforms for surveillance, rescue and search in confined environments due to their high flexibility and superior mobility. In such missions, it is essential that agents be able to explore and navigate autonomously in various environments. UAVs are capable of performing a mission more or less autonomously (Mostafa et al, 2017). Autonomous navigation in complex environments gives rise to engineering challenges. These problems require an integrated approach for perception, estimation, planning, control and high-level situational awareness.
Research on UAVs has deserved increasing interest in recent years, due to the variety of tasks these robots can perform. UAVs will continue to be applied in various areas such as traffic monitoring (Wang et al, 2015), logistics (Murray & Chu, 2015), road construction and urban development (Siebert & Teizer, 2014), integration to smart cities and cooperative networks (Mohammed et al, 2014), Safety and security (European Aviation Safety Agency, 2016), agriculture (Puri et al, 2017), search and rescue in disasters management (Erdelj et al, 2017), and many other applications. Its use is increasing in most areas due to their ease of deployment, high mobility and hovering ability.
Soft computing techniques are known as the essential tools capable for developing intelligent machines and solving nonlinear and mathematically system problems. Their applications include the design of intelligent autonomous systems/controllers and handling of complex systems with unknown parameters such as prediction of world economy, industrial process control and prediction of geological changes within the earth ecosystems. These paradigms have shown an ability to process information, adapt to changing environmental conditions, and learn from the environment.
The major soft computing techniques are following.
- (a)
Neural networks (NNs) modeling has a history dating back to the1950s,but only received an increasing amount of attention in the 1980s, with the emergence of the back-propagation algorithm (Rumelhart & McClelland, 1986), and have gained widespread recognition and approval. Their ability to learn and approximate functions has been the subject of considerable researchers’ interest and scientific inquiry. One only has to look at the many related industrial applications from the 1990s onwards and consult the abundant scientific literature on the subject to be convinced of this.
- (b)
Fuzzy logic (FL) introduced by Zadeh (Zadeh, 1965) in the 1960s, is a powerful tool that permits the treatment of vague, imprecise, uncertain and ill-defined knowledge and concept. Its use in the field of control (fuzzy control) was one of the first applications of this theory in industry with the work of Mamdani and Assilian (Mamdani & Assilian, 1975). Since then, the applications of FL have increased greatly to reach very diverse fields.
The use of fuzzy control (Milanés et al, 2012; Mon & Lin, 2012; David et al, 2013; Pinto et al, 2013; Ibrahim et al, 2014) is particularly interesting when there is no precise or even non-existent mathematical model of the system to be controlled or when the latter has strong non-linearity. Unlike traditional approaches to automation, which are largely based on a mathematical model, fuzzy logic control is based on a collection of linguistic rules in the form “If … Then” that reflect a human operator's control strategy.