A Receding Horizon Framework for Autonomy in Unmanned Vehicles

TL;DR

This paper introduces a unified receding horizon framework for autonomous unmanned vehicles, integrating guidance, navigation, and path-planning using model predictive control and moving horizon estimation for optimal, feasible operation.

Contribution

It presents a novel integrated framework combining guidance, navigation, and path-planning with receding horizon techniques for unmanned vehicle autonomy.

Findings

Successful path following in simulation with small errors

Effective obstacle avoidance demonstrated

Framework ensures feasible and optimal control solutions

Abstract

In this article we present a unified framework based on receding horizon techniques that can be used to design the three tasks (guidance, navigation and path-planning) which are involved in the autonomy of unmanned vehicles. This tasks are solved using model predictive control and moving horizon estimation techniques, which allows us to include physical and dynamical constraints at the design stage, thus leading to optimal and feasible results. In order to demonstrate the capabilities of the proposed framework, we have used Gazebo simulator in order to drive a Jackal unmanned ground vehicle (UGV) along a desired path computed by the path-planning task. The results we have obtained are successful as the estimation and guidance errors are small and the Jackal UGV is able to follow the desired path satisfactorily and it is also capable to avoid the obstacles which are in its way.