Dynamic Task Allocation and Path Planning for Multi-Aircraft Missions
View/ Open
Date
2024Author
Candemir, Doğan
xmlui.dri2xhtml.METS-1.0.item-emb
Acik erisimxmlui.mirage2.itemSummaryView.MetaData
Show full item recordAbstract
Military aircraft are advanced vehicles capable of performing various missions, primarily in air-to-air and air-to-ground scenarios. Before takeoff, a mission plan is prepared by ground crews, and pilots are expected to follow the planned route during flight. Adherence to the mission plan is critical, especially for combat aircraft, for the mission's success and the pilots' safety and health. However, due to unknown threats and dynamic environmental changes during flight, it may only sometimes be possible to follow the pre-planned route. Especially in missions involving multiple aircraft, a new plan must be prepared for each aircraft. In such situations, the leader pilot in charge of the mission is expected to assign tasks to other pilots and update the route for each aircraft accordingly. This thesis focuses on this problem and presents a study on dynamic task assignment and route planning to be used when the pre-prepared mission plan becomes invalid.
To tackle the challenges posed by both single and multiple aircraft missions, we have developed a new approach called Narrowed Regions-based Bidirectional Rapidly Exploring Random Tree (Narrowed-BiRRT). This method involves a matrix-based target assignment process for dynamic task allocation, followed by steps for route planning for the aircraft. The obtained routes are then optimized using a height optimization algorithm to prevent sudden altitude changes. This enhances the traceability of the routes generated by our developed method, making them more likely to be followed by real aircraft.
We tested the developed methods in scenarios, beginning with single-aircraft missions and progressively escalating threat levels in multi-aircraft situations. The algorithm showcased a 100\% convergence rate in all test scenarios, highlighting its capability to generate routes in any environment where a solution was identified. Our method was tested in a real scenario involving two aircraft and three threats, producing an optimal route of up to 10 km for each aircraft in a total of 1.4 seconds according to RRT, RRT* and RRT-Connect.
As a result, the methods developed in this thesis consistently produce optimal, environmentally responsive, threat-resistant, and adaptable routes tailored to different systems. The presented methodology offers practical solutions for single and multiple aircraft missions, making it applicable in autonomous flight systems for military and potentially civilian applications.