Abstract:This paper introduces a control architecture for real-time and onboard control of Unmanned Aerial Vehicles (UAVs) in environments with obstacles using the Model Predictive Path Integral (MPPI) methodology. MPPI allows the use of the full nonlinear model of UAV dynamics and a more general cost function at the cost of a high computational demand. To run the controller in real-time, the sampling-based optimization is performed in parallel on a graphics processing unit onboard the UAV. We propose an approach to the simulation of the nonlinear system which respects low-level constraints, while also able to dynamically handle obstacle avoidance, and prove that our methods are able to run in real-time without the need for external computers. The MPPI controller is compared to MPC and SE(3) controllers on the reference tracking task, showing a comparable performance. We demonstrate the viability of the proposed method in multiple simulation and real-world experiments, tracking a reference at up to 44 km/h and acceleration close to 20 m/s^2, while still being able to avoid obstacles. To the best of our knowledge, this is the first method to demonstrate an MPPI-based approach in real flight.

What problem does this paper attempt to address?

The paper primarily addresses the issue of real-time agile control of unmanned aerial vehicles (UAVs, commonly known as drones) in complex environments (containing obstacles). Specifically, the paper proposes a control architecture based on the Model Predictive Path Integral (MPPI) method to achieve safe navigation of drones during high-speed, flexible flight in environments with obstacles. The key contributions of the paper include: 1. **Design of the MPPI Controller**: By using the MPPI method, the controller can utilize the full nonlinear dynamics model of the drone and support more general cost functions, despite the higher computational demands. 2. **Parallel Optimization Strategy**: To achieve real-time control, the researchers adopted a parallel sampling optimization technique based on the Graphics Processing Unit (GPU), ensuring that the controller can run on onboard devices. 3. **Low-Level Constraints and Obstacle Avoidance**: A scheme for simulating nonlinear systems is proposed, which considers low-level constraints (such as motor performance limits) and can dynamically handle obstacle avoidance issues. 4. **Validation and Comparison**: Through multiple simulations and real-world experiments, the proposed MPPI controller is demonstrated to have comparable performance to Model Predictive Control (MPC) and SE(3) controllers in reference trajectory tracking tasks. Experiments show that the method can effectively avoid obstacles even at speeds up to 44 kilometers per hour and accelerations close to 20 meters per second squared. 5. **First Practical Flight Demonstration**: To the authors' knowledge, this is the first research work to demonstrate the MPPI method in actual flight, with all computations completed on the onboard platform. In summary, this research aims to overcome the challenges faced by traditional autonomous flight pipelines in handling obstacles, dynamic constraints, and target assignments through a novel control strategy, achieving efficient and safe flight of drones in unknown or partially known environments.

Model Predictive Path Integral Control for Agile Unmanned Aerial Vehicles

A Model Predictive Path Integral Method for Fast, Proactive, and Uncertainty-Aware UAV Planning in Cluttered Environments

Integrated Path Planning and Trajectory Tracking Control for Quadrotor UAVs with Obstacle Avoidance in the Presence of Environmental and Systematic Uncertainties: Theory and Experiment

Custom Non-Linear Model Predictive Control for Obstacle Avoidance in Indoor and Outdoor Environments

Real-time Motion Planning and Trajectory Tracking in Complex Environments Based on Bézier Curves and Nonlinear MPC Controller

Path Planning Based on Model Predictive Control Algorithm under Moving Threat

Enhancing AUV Autonomy With Model Predictive Path Integral Control

Nonlinear Model Predictive Control on SE(3) for Quadrotor Aggressive Maneuvers

Towards Efficient MPPI Trajectory Generation with Unscented Guidance: U-MPPI Control Strategy

Model Predictive Contouring Control with Barrier and Lyapunov Functions for Stable Path-Following in UAV systems

Neural Network Based Model Predictive Control for a Quadrotor UAV

Computationally Efficient Data-Driven MPC for Agile Quadrotor Flight

Real-time Planning of Minimum-time Trajectories for Agile UAV Flight

Performance, Precision, and Payloads: Adaptive Nonlinear MPC for Quadrotors

A Distributed Autonomous System for Multi-UAVs with Limited Visualization: Employing Dual-Horizon NMPC Controller

A Robust Disturbance-Rejection Controller Using Model Predictive Control for Quadrotor UAV in Tracking Aggressive Trajectory

Model Predictive Control for Micro Aerial Vehicle Systems (MAV) Systems

Online Predictive Visual Servo Control for Constrained Target Tracking of Fixed-Wing Unmanned Aerial Vehicles

Aerial navigation in obstructed environments with embedded nonlinear model predictive control

MPCC++: Model Predictive Contouring Control for Time-Optimal Flight with Safety Constraints

Experimental Validation of Linear and Nonlinear MPC on an Articulated Unmanned Ground Vehicle