GUPT Aet al.: CURVITRACK: CURVILINEAR TRAJECTORY TRACKING FOR HIGH-SPEED CHASE OF A USV 3MPC Solver Fast Fourier Transform USV Motion Prediction Setpoint Generator UA V Model Reference Tracker Position/Attitude Controller Vision-based Detector Attitude rate Controller IMU UA V Actuators Onboard Sensors State Estimator Odometry & Localisation ˆ x [ b w] n = 1 ..M p r d, η d ˆ r d, ˆ η d χ d 100 Hz ω d T d 100 Hz a d τ d 1 kHz x 100 Hz initialisation only x, R, ω 100 Hz R, ω b UA V plant Pixhawk autopilot MPC Architecture USV Prediction Model UA V Prediction ModelFigure 1: The entire UA V control architecture; the MPC landing controller (red block) is integrated into the MRS system [20] (grey blocks) and supplies the desired reference (velocity r d = null x y z null T and heading rate η d). In the MRS system, the first layer containing a Reference tracker processes the desired reference and gives a full-state reference χ to the attitude controller. The feedback Position/Attitude controller produces the desired thrust and angular velocities ( T d, ω d) for the Pixhawk flight controller (Attitude rate controller). The State estimator fuses data from Odometry & localisation methods to create an estimate of the UA V translation and rotation ( x, R). The Vision-based Detector obtains the visual data from the camera and sends the pose information b of the USV to the MPC. The individual states are sent to their respective prediction models, and using these predictions, the MPC generates the desired control reference according to the cost function.

We address the challenge of real-time planning of minimum-time trajectories over multiple waypoints, onboard multirotor UAVs. Previous works demonstrated that achieving a truly time-optimal trajectory is computationally too demanding to enable frequent replanning during agile flight, especially on less powerful flight computers. Our approach overcomes this stumbling block by utilizing a point-mass model with a novel iterative thrust decomposition algorithm, enabling the UAV to use all of its collective thrust, something previous point-mass approaches could not achieve. The approach enables gravity and drag modeling integration, significantly reducing tracking errors in high-speed trajectories, which is proven through an ablation study. When combined with a new multi-waypoint optimization algorithm, which uses a gradient-based method to converge to optimal velocities in waypoints, the proposed method generates minimum-time multi-waypoint trajectories within milliseconds. The proposed approach, which we provide as open-source package, is validated both in simulation and in real-world, using Nonlinear Model Predictive Control. With accelerations of up to 3.5g and speeds over 100 km/h, trajectories generated by the proposed method yield similar or even smaller tracking errors than the trajectories generated for a full multirotor model.

The tf$distribute$Strategy API provides an abstraction for distributing your training across multiple processing units. The goal is to allow users to enable distributed training using existing models and training code, with minimal changes. This tutorial uses the tf$distribute$MirroredStrategy, which does in-graph replication with synchronous training on many GPUs on one machine. Then, it uses all-reduce to combine the gradients from all processors and applies the combined value to all copies of the model. MirroredStategy is one of several distribution strategy available in TensorFlow core.