In this work, a motion planner for small fixed-wing unmanned aerial vehicles (UAVs) is presented. The planner is designed for motion in three dimensions (3d) and six degrees of freedom (6D0F) with constraints and obstacles. It is set up with Nonlinear Model Predictive Control (NMPC) and is especially developed for high-frequent onboard execution in real-time. Furthermore, a new implementation method of an NMPC on a Field Programmable Gate Array (FPGA) is established. The idea is to utilize FPGAs for parallel numerical model prediction in hardware. This new implementation method and a suited NMPC setup allow for efficient implementation of the NMPC planning algorithm on a relatively small onboard computer with a FPGA module with low power consumption. Finally, this enables the application of the motion planner to a small unmanned fixed-wing UAV in flight. First, an efficient setup of the NMPC motion planner is described. The associated optimization problem is solved with numerical time-domain simulations with a set of predefined candidate control options. In order to evaluate these control options, the nonlinear 3d kinematics is used as a prediction model in combination with actions motivated from flight mechanics. Due to the control structure, the planning problem can be solved even with a low number of candidate control options. Furthermore, the computational effort is reduced by exploiting the results of a nonlinear controllability analysis. Next, the applicability of NMPC with finite prediction horizon is studied in the context of this work with a stability analysis without stability constraints or terminal penalty terms. The objective here is to find a prediction horizon with a low computational load that still ensures closed loop stability for the system. A new method for the implementation of NMPC on FPGAs is introduced. In this method, hardware logic elements of a FPGA are used for parallel numerical time-domain simulations. Candidate control options can thus be evaluated with low and guaranteed computation time. Through this method and an integrated consideration of the required hardware logic elements, available computation time, and numerical precision, NMPC can be executed efficiently in real-time on FPGA modules, which are especially suitable for small fixed-wing UAVs with stringent mass limitations. The setup of the NMPC planner together with the suited implementation method finally enables the high-frequent execution of the NMPC planner (computation time approx. 5.1 ms) on a low-power onboard computer. Flight tests with a small fixed-wing UAV and hardware-in-the-loop real-time tests verify the NMPC setup and the FPGA implementation method in scenarios with attitude constraints, obstacles, and even in the presence of wind.


    Zugriff

    Zugriff über TIB

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Real-time predictive motion planning for fixed-wing aerial vehicles


    Beteiligte:


    Erscheinungsdatum :

    2014


    Format / Umfang :

    122 Seiten, Bilder, Tabellen, 100 Quellen




    Medientyp :

    Hochschulschrift


    Format :

    Print


    Sprache :

    Englisch





    Dynamic motion planning for aerial surveillance on a fixed-wing UAV

    Darbari, Vaibhav / Gupta, Saksham / Verma, Om Prakash | IEEE | 2017


    Wing strike reduction for small fixed wing uncrewed aerial vehicles

    Henry, Todd C. / Hrynuk, John T. / Del Colliano, Adam et al. | Elsevier | 2023


    Delivering Newspapers Using Fixed Wing Unmanned Aerial Vehicles

    Agarwal, Varun / Tewari, Rajiv Ranjan | Springer Verlag | 2021


    Flocking Algorithm for Fixed-Wing Unmanned Aerial Vehicles

    Kownacki, Cezary / Ołdziej, Daniel | Springer Verlag | 2015