This letter presents a perception-aware and motorlevel non-linear model predictive control scheme for multi-rotor aerial vehicles. Our formulation considers both real actuation limitations of the platform, and realistic perception objectives for the visibility coverage of an environmental feature while performing a reference task. It directly produces the rotor-level (torque) inputs of the platform motors at high frequency, hence it does not require an intermediate unconstrained controller to work. It is also meant to be generic, by covering standard coplanar quadrotors as well as tilted-propeller multi-rotors. We propose an open-source fully onboard implementation of the method, capable of running at 500 Hz under the intermittent and noisy measurements of one or more cameras. The implementation is extensively tested both in simulation and in real experiments with two substantially different multi-rotor platforms, an underactuated and a fully actuated one, both equipped with two cameras, clearly demonstrating the practicability and high performance of the method.

No abstract

In this paper, we present a Perception-constrained Nonlinear Model Predictive Control (NMPC) framework for the real-time control of multi-rotor aerial vehicles. Our formulation considers both constraints from a perceptive sensor and realistic actuator limitations that are the rotor minimum and maximum speeds and accelerations. The formulation is meant to be generic and considers a large range of multi-rotor platforms (such as underactuated quadrotors or tilted-propellers hexarotors) since it does not rely on differential flatness for the dynamical equations, and a broad range of sensors, such as cameras, lidars, etc... The perceptive constraints are expressed to maintain visibility of a feature point in the sensor's field of view, while performing a reference maneuver. We demonstrate both in simulation and real experiments that our framework is able to exploit the full capabilities of the multi-rotor, to achieve the motion under the aforementioned constraints, and control in real-time the platform at a motor-torque level, avoiding the use of an intermediate unconstrained trajectory tracker.

In this article, we consider the problem of delivering an object to a human coworker by means of an aerial robot.To this aim, we present an ergonomics-aware Nonlinear Model Predictive Control (NMPC) designed to autonomously perform the handover. The method is general enough to be applied to any mobile robot with a minimal adaptation of the robot model. Our formulation lets the NMPC steer the robot toward a handover location optimizing the human coworker ergonomics metrics, which includes the predicted joint torques of the human. The motion task is expressed in a frame relative to the human, whose motion model is included in the equations of the NMPC. This allows the controller to reactively adapt to the human movements by predicting her future poses over the horizon. The control framework also accounts for the problem of maintaining visibility on the human coworker, while respecting both the actuation and state limits. A safety barrier is also embedded in the controller to avoid any risk of collision with the human partner. Realistic simulations are used to validate the feasibility of the approach and the source code of the implementation is released open-source.