A 3D navigation algorithm switching between waypoint and Bezier curves based local plans for micro air vehicles

Abstract

Micro Air Vehicles (MAVs) are playing an increasingly prominent role in our lives. Numerous studies have been conducted on autonomous mobility. In this study, an architecture has been developed for 3D mapping and 3D navigation for MAVs. The flight time of a MAV is considerably shorter than ground vehicles. Therefore, they are expected to perform tasks more efficiently and with higher performance. MAVs can plan a route to a given target and follow this path using the developed 3D navigation algorithm. Tracking paths produced by classical waypoint-based planners is slow. For this reason, a local path planner based on Bezier curves has been developed for MAVs, with path planning represented as a curve on the low-level control card. Since it is not always possible to reach the given target with a single curve, the multi-hop Bezier curves approach has been proposed. Particularly in environments with low complexity, this approach yields very effective results. In cases where the Bezier path cannot be generated in high-complexity environments, a metric has been developed to measure the complexity, allowing for a switch between waypoint and multi-hop Bezier curves based local plans on the environment complexity metric. Thanks to these developed methods, the aim is to extend the flight time of MAVs by optimizing their battery usage. Within the scope of this study, all experiments were conducted in the ROS-GAZEBO simulation environment. Firstly, waypoint-based local planning methods such as Dijkstra and A*, which will function within the navigation algorithm, were implemented to operate in 3D, and performance tests were conducted in the designed simulation environments. Subsequently, a multi-hop Bezier curves-based local planner was developed, and performance comparisons were made with waypoint-based methods using the same simulation environments, taking into account environmental complexity parameters.