Optimal Motion Planning in GPS-Denied Environments Using Nonlinear Model Predictive Horizon

Navigating robotic systems autonomously through unknown, dynamic and GPS-denied environments is a challenging task. One requirement of this is a path planner which provides safe trajectories in real-world conditions such as nonlinear vehicle dynamics, real-time computation requirements, complex 3D environments, and moving obstacles. This paper presents a methodological motion planning approach which integrates a novel local path planning approach with a graph-based planner to enable an autonomous vehicle (here a drone) to navigate through GPS-denied subterranean environments. The local path planning approach is based on a recently proposed method by the authors called Nonlinear Model Predictive Horizon (NMPH). The NMPH formulation employs a copy of the plant dynamics model (here a nonlinear system model of the drone) plus a feedback linearization control law to generate feasible, optimal, smooth and collision-free paths while respecting the dynamics of the vehicle, supporting dynamic obstacles and operating in real time. This design is augmented with computationally efficient algorithms for global path planning and dynamic obstacle mapping and avoidance. The overall design is tested in several simulations and a preliminary real flight test in unexplored GPS-denied environments to demonstrate its capabilities and evaluate its performance.

Automated summary

The study introduces a method that combines local path planning with a graph-based planner to enable autonomous navigation of unmanned vehicles in GPS-denied subterranean environments. By utilizing the Nonlinear Model Predictive Horizon method, feasible, optimal, smooth, and collision-free paths are generated. The design also includes computationally efficient algorithms for global path planning and dynamic obstacle mapping and avoidance.