查看更多>>摘要：Thruster control allocation (TCA) is a key functionality for many spacecraft, with a significant impact on control performance, propellant consumption, and fault tolerance. Propellant-optimal solutions are desirable and are either based on onboard numerical optimization, or explicit optimization via the use of offline-generated look-up tables (LUTs). This article proposes a TCA and modulation method of the latter type by using multiparametric programming and presents a novel fast LUT evaluation algorithm. Fault tolerance and the handling of non-attainable control commands with full controllability exploitation are also addressed. Furthermore, the solution is extended to include the non-convex minimum impulse bit (MIB) constraint, where the proposed solution can find the global optimum. The use of this constraint is demonstrated in a close-range orbital rendezvous scenario, yielding significant improvements to the performance of boosts, forced motions, and station-keeping maneuvers, at the cost of greater propellant consumption and computation time. Results in consumer hardware for a12-thruster configuration show a worst case onboard computation time of $7~\mu $ s and 0.5 ms for the cases without and with the MIB constraint, which are up to two orders of magnitude lower than those for numerical optimization with a state-of-the-art optimizer. The proposed onboard algorithms are simple, non-iterative, and have worst case computational effort guarantees.
查看更多>>摘要：In this article, we address the model predictive control (MPC) problem for continuous-time linear time-invariant systems, with both state and input constraints. For computational efficiency, existing approaches typically discretize both dynamics and constraints, which potentially leads to constraint violations in between discrete-time instants. In contrast, to ensure strict constraint satisfaction, we equivalently replace the differential equations with linear mappings between state, input, and flat output, leveraging the differential flatness property of linear systems. By parameterizing the flat output with piecewise polynomials and employing Markov-Lukács theorem, the original MPC problem is then transformed into a semidefinite programming (SDP) problem, which guarantees the strict constraints satisfaction at all time. Furthermore, exploiting the fact that the proposed SDP contains numerous small-sized positive semidefinite (PSD) matrices as optimization variables, we propose a primal-dual hybrid gradient (PDHG) algorithm that can be efficiently parallelized, expediting the optimization procedure with GPU parallel computing. The simulation and experimental results demonstrate that our approach guarantees rigorous adherence to constraints at all time, and our solver exhibits superior computational speed compared to existing solvers for the proposed SDP problem.
查看更多>>摘要：This article presents a robust feedforward design approach using hybrid modeling to improve the output tracking performance of feed drives. Geared toward the use for feedforward design, the hybrid model represents the dominant linear dynamics with a flat analytical model and captures the output nonlinearity by Gaussian process (GP) regression. The feedforward control is based on the model inversion, and the design procedure is formulated as a signal-based robust control problem, considering multiple performance objectives of tracking, disturbance rejection, and input reduction under uncertainties. In addition, the technique of structured $\mu $ synthesis is applied, which allows direct robust tuning of the fixed-structure feedforward gains and ensures the applicability in industrial hardware. The proposed methodological approach covers the entire procedure from modeling to control architecture selection and weights design, delivering an end-to-end strategy that accounts for performance and robustness requirements. Validated on an industrial milling machine with real-time capability, the proposed robust controller reduces the mean absolute tracking error in the transient phase by 83% and 63% compared to the industrial standard baseline feedforward and the nominal design, respectively. Even with a variation of 20% in the model parameters, the robust feedforward still reduces the error by 58% in the worst case with respect to the baseline.
Daksh ShuklaHady BenyamenShawn KeshmiriNicole M. Beckage...
872-886页
查看更多>>摘要：A significant challenge in designing flight controllers lies in their dependency on the quality of dynamic models. This research explores the potential of artificial intelligence-based flight controllers to generalize control actions around policies rather than relying solely on the accuracy of dynamic models. An engineering-level, low-fidelity, linearized model of fixed-wing uncrewed aircraft is used to train a multi-input multi-output (MIMO) flight controller, employing the deep deterministic policy gradients (DDPG) algorithm, to maintain cruise velocity and altitude. While existing literature often concentrates on simulation-based assessments of reinforcement learning (RL)-based flight controllers, this research employs an extensive flight test campaign including 15 flight tests to explore the reliability, robustness, and generalization capability of RL algorithms in tasks they were not specifically trained for, such as changing cruise altitude and velocity. The RL controller outperformed a well-tuned linear quadratic regulator (LQR) on several control tasks. Furthermore, a modification in the DDPG algorithm is presented to enhance the ability of RL controllers to evolve through experience gained from actual flights. The evolved controllers present different behavior compared to the original controller. Comparative flight tests underscored the crucial role of the ratio of actual flight data to the number of simulation-based training instances in optimizing the evolved controllers.
Maxime GrossoPierre RiedingerJamal DaafouzSerge Pierfederici...
887-902页
查看更多>>摘要：In light of recent advances in harmonic modeling and control theory, we present a novel approach for controlling a three-phase grid-tied ac-dc converter. Our methodology involves deriving an infinite-dimensional harmonic bilinear model and employing forwarding control techniques to design globally stable state feedback. This framework enables the integration of control objectives tailored to harmonic distortion, mitigation, and the tracking of periodic trajectories, leveraging integral actions. Moreover, we demonstrate the transposition of our harmonic control design into a periodic nonlinear control scheme in the time domain, preserving the established stability guarantees. These assurances stem from recent fundamental theoretical findings, outlining the conditions for a bijection between harmonic- and time-domain trajectories. Through illustrative simulations and experimental setups, we assess the methodology’s effectiveness in real-time total harmonic distortion (THD) reduction and rejection of specific harmonic disturbances. The obtained results affirm the successful accomplishment of our control objectives, validating the efficacy of the proposed harmonic control approach.
查看更多>>摘要：This article introduces a multilayer navigation algorithm for a fleet of unmanned aerial vehicles (UAVs). The proposed architecture consists of a fusion of virtual point controllers and potential field techniques. On the one hand, a potential function is constructed for every agent such that its position smoothly and robustly converges to a virtual guidance point while avoiding collisions with other agents. The virtual points, on the other hand, are controlled to fulfill a swarm control goal such as target tracking, station keeping, or search and rescue missions. Therefore, the suggested system has two levels of hierarchy, but the algorithm can be generalized for multiple levels. The vehicle translational and rotational dynamics are controlled using an internal loop based on gradient tracking and sliding mode controllers. The architecture is validated in simulations and real-time experiments, showing good performance for the closed-loop system.
查看更多>>摘要：The intrinsic hysteresis nonlinearity of piezo-actuated stages (piezo stages) poses a significant challenge for precise trajectory tracking at high speeds. In response, we propose a deep parallel (dPara) model that effectively captures the dynamics of the piezo stage using historical voltage–displacement data over a concise time period. The dPara model, incorporating a parallel combination of a linear block and a feedforward neural network (FNN), exhibits exceptional performance with relative prediction errors ranging between 0.10% and 0.18% on sinusoidal trajectories at frequencies up to 72% of the resonance frequency of the piezo stage. By leveraging this parallel structure, we adapt the reference trajectory for a complex nonlinear model predictive control (MPC), leading to the development of the reference-adaptation MPC (RA-MPC). Furthermore, we design a coordinate ascent algorithm to solve the quadratic programming (QP) problem derived from the RA-MPC at a high frequency of 10 kHz. To assess the superiority of the proposed RA-MPC, comprehensive experiments are conducted under sinusoid, sawtooth, and staircase reference trajectories. Notably, it achieves maximum tracking errors (MTEs) ranging from 0.0263 to $0.7136 \; \mu $ m for desired speeds spanning from 40 to $20\,000 \; \mu $ m/s.
查看更多>>摘要：This article presents a robot control algorithm suitable for safe reactive navigation tasks in cluttered environments. The proposed approach consists of transforming the robot workspace into the ball world, an artificial representation where all obstacle regions are closed balls. Starting from a polyhedral representation of obstacles in the environment, obtained using exteroceptive sensor readings, a computationally efficient mapping to ball-shaped obstacles is constructed using quasi-conformal (QC) mappings and Möbius transformations. The geometry of the ball world is amenable to provably safe navigation tasks achieved via control barrier functions (CBFs) employed to ensure collision-free robot motions with guarantees both on safety and on the absence of deadlocks. The performance of the proposed navigation algorithm is showcased and analyzed via extensive simulations and experiments performed using different types of robotic systems, including manipulators and mobile robots.
查看更多>>摘要：For safe navigation in dynamic uncertain environments, robotic systems rely on the perception and prediction of other agents. Particularly, in occluded areas, where cameras and light detection and ranging (LiDAR) give no data, the robot must be able to reason about the potential movements of invisible dynamic agents. This work presents a provably safe motion planning scheme for real-time navigation in an a priori unmapped environment, where occluded dynamic agents are present. Safety guarantees are provided based on the reachability analysis. Forward reachable sets associated with potential occluded agents, such as pedestrians, are computed and incorporated into planning. An iterative optimization-based planner is presented that alternates between two optimizations: nonlinear model predictive control (NMPC) and collision avoidance. The recursive feasibility of the MPC is guaranteed by introducing a terminal stopping constraint. The effectiveness of the proposed algorithm is demonstrated through simulation studies and hardware experiments with a TurtleBot robot equipped with a LiDAR system. The video of experimental results is also available at: https://youtu.be/OUnkB5Feyuk.
NETL
NSTL
IEEE