A new method based on motion primitives to compute 3D path planning close to helicopters' flight dynamics limits

Abstract

3D path planning close to helicopters' flight dynamics limits is a critical challenge to enhance missions' assistance or autonomy. Pure dynamics model with successive integrations, Pythagorian Hodograph methods or fictive potential fields bring different drawbacks that cannot solve the problem of connecting 3D positions with associated complex set of constraints (speed orientation, curvature, torsion...) in a satisfying way. We propose in the present paper a new method based on enhanced 3D motions primitives. Their construction has been possible due to the transformation of flight loop numerical simulations into a symbolic and infinitely derivable model based on Beziers curves. Our algorithm defines strategic forms like cylinders referring to Dubins circles or vertical planes and hooks them up with the ad hoc 3D motion primitives while minimizing the curvilinear abscissa according to new principles, which are introduced. The concepts are tested for two path-planning problems and then validated against a reconstructed helicopter accident. The performance of the algorithm is measured versus an ideal Dubins-like 3D path metrics. The extremely good results we obtained pushes the challenge a step further to build up “an all flight contexts” model for the future.