Autonomous landing of unmanned aerial vehicle with capability to avoid reactive collision using a mixed guidance scheme, which is a combination of ‘geometric optimization technique’ and ‘collision cone approach’ has been presented in this paper. Geometric optimization technique ensures minimal deviation from nominal trajectory to achieve optimal velocity, heading and elevation. Collision cone approach predicts collision ahead of time and provides aiming point for vehicle towards conflict resolution. A sobolev norm based robust neuro adaptive controller is designed to control the autonomous landing of vehicle with capability to avoid reactive collision under unknown external disturbances. A multilayer feed-forward network is designed based on radial basis function to estimate the unknown disturbances by fast learning. The autonomous landing vehicle senses the other UAV approaching towards it through its on board stereo vision sensing and performs a collision avoidance manoeuvre if the minimum predicted separation between UAVs is less than a predefined safety threshold distance. Both vehicles are simulated with a full fledged six degree of freedom model of a real vehicle.
Autonomous Landing of UAVs with Reactive Collision Avoidance using Mixed Guidance Scheme and Neuro Adaptive Controller
20.09.2021
1736659 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Nonlinear Geometric and Differential Geometric Guidance of UAVs for Reactive Collision Avoidance
British Library Conference Proceedings | 2010
|Reactive collision avoidance of multiple realistic UAVs
Online Contents | 2011
|Reactive collision avoidance of multiple realistic UAVs
Emerald Group Publishing | 2011
|