Embedded Hierarchical MPC for Autonomous Navigation
Dennis Benders, Johannes Köhler, Thijs Niesten, Robert Babu�ka, Javier Alonso-Mora, Laura Ferranti
AI summary
Problem
Single-layer nonlinear MPC faces a computational trade-off on embedded hardware between long-term planning and fast tracking, often forcing compromises in maneuverability or safety. Existing hierarchical methods typically linearize dynamics or rely on pre-defined motion primitives, limiting their applicability to complex nonlinear systems.
Approach
The framework codesigns a slow-rate planning MPC for long-horizon trajectory generation and a fast-rate tracking MPC for precise execution, both using the same nonlinear model. Constraint tightening and offline terminal design guarantee that the tracker can reliably follow the plan while avoiding obstacles.
Key results
- Formal proof of collision avoidance and recursive feasibility
- Real-time implementation on a quadrotor’s embedded NVIDIA Jetson computer
- Fivefold increase in planning horizon compared to single-layer MPC
- Faster goal-reaching and improved altitude maintenance in simulations and experiments
Why it matters
Provides a practical, theoretically guaranteed pathway for deploying advanced nonlinear MPC on resource-constrained mobile robots.
Abstract
To efficiently deploy robotic systems in society, mo- bile robots must move autonomously and safely through complex environments. Nonlinear model predictive control (MPC) methods provide a natural way to find a dynamically feasible trajectory through the environment without colliding with nearby obstacles. However, the limited computation power available on typical em- bedded robotic systems, such as quadrotors, poses a challenge to running MPC in real time, including its most expensive tasks: constraints generation and optimization. To address this problem, we propose a novel hierarchical MPC scheme that consists of a planning and a tracking layer. The planner constructs a trajectory with a long prediction horizon at a slow rate, while the tracker ensures trajectory tracking at a relatively fast rate. We prove that the proposed framework avoids collisions and is recursively feasi- ble. Furthermore, we demonstrate its effectiveness in simulations and lab experiments with a quadrotor that needs to reach a goal position in a complex static environment. The code is efficiently implemented on the quadrotor’s embedded computer to ensure real-time feasibility. Compared to a state-of-the-art single-layer MPC formulation, this allows us to increase the planning horizon by a factor of 5, which results in significantly better performance.