Recent advances in aerial robotics have enabled the use of multirotor vehicles for autonomous payload transportation. Resorting only to classical methods to reliably model a quadrotor carrying a cable-slung load poses significant challenges. On the other hand, purely data-driven learning methods do not comply by design with the problem’s physical constraints, especially in states that are not densely represented in training data. In this work, we explore the use of physics-informed neural networks to learn an end-to-end model of the multirotor-slung-load system and, at a given time, estimate a sequence of the future system states. An LSTM encoder-decoder with an attention mechanism is used to capture the dynamics of the system. To guarantee the cohesiveness between the multiple predicted states of the system, we propose the use of a physics-based term in the loss function, which includes a discretized physical model derived from first principles together with slack variables that allow for a small mismatch between expected and predicted values. To train the model, a dataset using a real-world quadrotor carrying a slung load was curated and is made available. Prediction results are presented and corroborate the feasibility of the approach. The proposed method outperforms both the first principles physical model and a comparable neural network model trained without the physics regularization proposed.

This paper presents the design, analysis, and experimental validation of a globally asymptotically stable (GAS) filter for simultaneous localization and mapping (SLAM), with application to unmanned aerial vehicles. The SLAM problem is formulated in a sensor-based framework and modified in such a way that the underlying system structure can be regarded as linear time varying for observability analysis and filter design purposes, from which a linear Kalman filter with GAS error dynamics follows naturally. The performance and consistency validation of the proposed sensor-based SLAM filter are successfully assessed with real data, acquired indoors, using an instrumented quadrotor.

This paper proposes an estimation algorithm for the determination of attitude installation matrix for laser detection and ranging systems (LADAR) mounted onboard autonomous vehicles, without requiring any prior knowledge on the terrain where the calibration mission is performed. The use of autonomous vehicles equipped with LADAR systems to conduct fully automatic surveys of terrain, infrastructures, or just to navigate safely in unknown environments, motivates the research on increasingly precise LADAR data acquisition and processing algorithms, for which the determination of the correct installation matrix is critical. The proposed methods rely on the minimization of the errors between several acquired data sets, by comparing each measured data set and a surface representation of the others. This error functional is minimized resorting to optimization tools for Riemannian manifolds enabling direct estimation of the installation matrix on the group of special orthogonal matrices SO(3). The proposed technique is extensively tested using simulated LADAR data sets under realistic acquisition conditions.

This paper proposes two estimation algorithms for the determination of attitude installation matrix for Laser Detection and Ranging systems (LADAR) mounted onboard autonomous vehicles. The use of autonomous vehicles equipped with LADAR systems to conduct fully automatic surveys of terrain, infrastructures, or just to navigate safely in unknown environments, motivates the research on increasingly precise LADAR data acquisition and processing algorithms, for which the determination of the correct installation matrix is critical. The proposed methods rely on the minimization of the errors between the measured data set and a representation of the real calibration surface. To minimize this error, two nonlinear optimization techniques are proposed, one that estimates the ZYX Euler angles and a second that uses optimization tools for Riemannian manifolds enabling direct estimation of the installation matrix on the group of special orthogonal matrices SO(3). The proposed techniques are extensively tested and their effectiveness compared resorting to simulated LADAR data sets under realistic noise conditions.