Developments in the aviation industry have led to an increase in air traffic, requiring all operations to be more flexible and efficient without disregarding safety standards. Simultaneously, the amount and variety of real-time data available to air transport systems and their crews are also increasing. This situation calls for higher levels of automation where humans in the loop focus on their supervising role. For this reason, a Smart Auto-Flight Control System (SAFCS) is being developed to continuously plan, execute and monitor airborne missions in which it is involved. It uses digital communications to interact with other agents and applies path planning algorithms to find the best trajectories for each scenario.
The Rapidly-exploring Random Tree (RRT) is one of such path planning algorithms, first introduced as being capable of dealing with high-dimensional, nonholonomic systems with kinodynamic constrains. Such characteristics have led to widespread applications and the continuous development of RRTs over the past years including extensions of these algorithms to satisfy anytime, dynamic and online requirements.
This dissertation integrates the RRT planning family into the SAFCS project, discussing its benefits and selecting the algorithms that better fit the requirements. In certain cases, enhancements to existing algorithms are proposed to solve specific problems. Finally, the implementation is evaluated and results are presente