Fault-tolerant predictive control with trajectory planning in the presence of obstacles

• Autor(a) principal: Rubens Junqueira Magalhães Afonso
• Data de Publicação: 2015
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações do ITA
• Texto Completo: http://www.bd.bibl.ita.br/tde_busca/arquivo.php?codArquivo=3212
• topic: Tolerância a falhas; Controle de voo; Aeronaves; Trajetórias de voo; Controle preditivo; Engenharia aeronáutica; Controle

This work addresses the problem of trajectory planning in the context of predictive control. Initially, planners in the literature using predictive control techniques with binary variables are extended to consider situations in which it is necessary that the vehicle visits multiple targets featuring a limited amount of fuel. A reward for visiting each target set is included to tune the compromise between saving fuel and visiting as many targets as possible. The second part brings contributions aimed at reducing computational complexity of the problem of planning and execution of trajectories. For this purpose, techniques for reducing the number of binary variables necessary to implement obstacle avoidance constraints are studied. In this context, a more economical procedure for encoding the problem is presented. Moreover, a method is proposed for generation of waypoints to be traversed by the vehicle in order to complete the mission. The resulting waypoints are passed to the control layer responsible for guiding the vehicle through them. The advantage of this dual-layer approach is that the online computational effort for determining the control signal is significantly reduced. With the approach of multiple sets and rewards, it is possible to accommodate a fault which causes a reduction in the amount of fuel available. Moreover, using the waypoint planning technique, the online computational load is reduced, making it viable to deal with problems involving more stringent control constraints possibly arising from actuator faults.