Aerial Vision Based Guidance and control for Perception-Less Ground Vehicle

The synergy between unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) within a heterogeneous system presents diverse avenues for cooperative operation. These collaborations leverage the complementary strengths of both systems, culminating in remarkable improvements in task execution efficiency. In this paper, we explore the concept of a perception-sensor-deprived UGV, and introduce a ground vehicle guidance and control system that exclusively relies on aerial perception to perform path-following tasks. The cooperative architecture is composed of two parts, one for detecting and tracking the ground trajectory using aerial images retrieved from the UAV. And the second for computing the control inputs for ground autonomous navigation. A nonlinear controller for the ground vehicle is designed using sliding mode with fractional components and its stability analysis is proved using the Lyapunov theory. The cooperative architecture was proved in realtime tests showing that experimental results underscore the well efficacy of this system in autonomously guiding a ground vehicle along a predefined path solely based on aerial imagery, opening new horizons for advanced robotics and autonomous systems. To our best of knowledge, this represents the inaugural endeavor in the development of a car-like robot dynamics controller based on non-stationary aerial imagery for path following utilizing a non-learning approach.