Trajectory Optimization and Replanning Framework for a Micro Air Vehicle in Cluttered Environments

Hyeonbeom Lee,Hoseong Seo,Hyeong-Geun Kim

doi:10.1109/access.2020.3011401

Abstract

In the present study, we propose a trajectory optimization and replanning algorithm for micro air vehicles (MAVs) in cluttered environments. To generate the path of an MAV in a cluttered environment, we first design an offline global path optimization algorithm. This algorithm generates a global trajectory for safe aerial delivery; this trajectory enables an MAV to avoid static obstacles marked in the navigation map and satisfies the MAV's initial and arrival velocities. The MAV's trajectory is replanned by exploiting dynamic movement primitives (DMPs) and a time adjustment algorithm to enable computationally efficient unknown obstacle avoidance in local path planning. To validate the applicability of the proposed algorithm, we compare simulation results with those obtained using an existing approach based on DMPs. Furthermore, an autonomous flight is demonstrated in an outdoor environment using a custom-made MAV driven by the proposed approach.

Highlights

Recent changes in social phenomena, such as an increase in the popularity of online shopping, have been driving the development of aerial robots for efficient logistics management
CONTRIBUTION we propose a framework for micro air vehicles (MAVs) that includes trajectory optimization using a static three-dimensional (3D) map and a real-time replanning algorithm in an unknown obstacle environment
Gradient-based optimization planning [7] can generate a path considering the dynamic constraints of robots. These methods [7], [17], [18] experience high computational loads; path planning exclusively based on Mixed-integer programming (MIP) or gradient-based optimization is difficult to apply to the problem of real-time path adjustment for dynamic or unknown obstacles that are not marked in maps

Summary

Introduction

Recent changes in social phenomena, such as an increase in the popularity of online shopping, have been driving the development of aerial robots for efficient logistics management. Assuming that the global waypoints are given, the global desired trajectory can be generated to perform safe aerial missions wherein the MAV can avoid static obstacles marked in the map and satisfy the initial and final velocity conditions. A. CONTRIBUTION we propose a framework for MAVs that includes trajectory optimization using a static three-dimensional (3D) map and a real-time replanning algorithm in an unknown obstacle environment.

Objectives

Results

Conclusion