Learning and extracting high-level features from point cloud is the key to improving the segmentation performances on point clouds for many networks. At present, many networks present very deep structures to extract high-level features for 3D perception. However, we argue that even better results can be achieved by (i) building feature vectors that integrates multi-scale geometric features, and (ii) exerting discriminative constraints on the learning of mid-levels features. In this paper, we propose a Multi-scale Neighborhood Feature Extraction and Aggregation Model (MNFEAM) to enhance feature extraction for point cloud learning. We try to first extract multi-scale neighborhood information for each input point and then aggregate local information of a mid-level locality feature space, and finally integrate the aggregated local and the global feature vectors. A new discriminative loss function is designed to strengthen the coarse semantics on mid-levels features so that the semantic abstraction process can be improved and accelerated. We improve the performances of three popular networks for point cloud segmentation using the proposed MNFEAM on standard 3D datasets.

This article investigates the network-based multiple operating points cooperative dynamic positioning of multiple unmanned surface vehicles. First, a dynamic positioning system model is established for multiple unmanned surface vehicles. Second, a network-based multiple operating points cooperative dynamic positioning control protocol and a resultant system model for multiple unmanned surface vehicles are constructed under fixed topologies. Third, a networked multiple operating points cooperative dynamic positioning control scheme, which can guarantee the multiple operating points cooperative dynamic positioning error exponentially converge to a bounded region, is proposed. The dynamic positioning control scheme design is then extended to the case of switching topologies. Finally, the multiple operating points cooperative dynamic positioning performance analysis verifies the efficiency of the proposed control schemes.

This article is concerned with the problem of dynamic event-triggered (DET) scaled consensus control for multi-agent systems (MASs) in reliable and unreliable networks. First, by introducing an auxiliary dynamic variable (ADV) obeying its own dynamics, a novel DET mechanism is designed. A nonzero exponentially decaying term is introduced in the DET scheme to enlarge the interval between two consecutive trigger instants. In reliable networks, the DET mechanism works with a set of fixed parameters for the nonzero exponentially decaying term. Second, by introducing a stochastic variable into the parameters to describe the uncertain denial-of-service (DoS) attacks, a resilient DET mechanism is proposed. In unreliable networks, the resilient DET mechanism adjusts the set of parameters between two sets of values by detecting the launch of DoS attacks. Then, by choosing suitable Lyapunov–Krasovskii functions (LKFs), some stability criteria are derived for scaled consensus of MASs both in reliable and unreliable networks. Based on the stability criteria expressed in matrix inequalities, a control design algorithm based on genetic algorithm (GA) is proposed to calculate the control gains and event trigger parameters jointly. Finally, the effectiveness of the results is verified via a numerical example of multiple two-wheel mobile vehicles.

This article is concerned with the event-triggered output feedback cluster consensus of leader–following multi-agent systems (MASs) under limited communication resources. Specifically, the distributed agents are divided into several clusters to accomplish different collective tasks under diverse intracluster and intercluster communications. First, to alleviate excessive communication resource consumption, two sampled-data-based event-triggered schemes are developed to distinguish agent-to-agent communications within clusters and between clusters. Based on these schemes, an event-based cluster consensus control protocol is proposed to solve the problem. Then, sufficient criteria on asymptotic stability of the resulting closed-loop system are derived and expressed in terms of matrix inequalities. It is noteworthy that the derived criteria for controller design are nonlinear and nonconvex with respect to the output feedback control gains and triggering parameters. To handle this issue, a modified genetic algorithm (MGA) with multiple subpopulations is proposed, where the subpopulations are independent of each other. The key feature of the designed MGA lies in that the fitness value is described as an accumulation of initial value and weighing value of each matrix inequality. Finally, an application of satellite formation flying is exemplified to demonstrate the effectiveness of the derived theoretical results.

This article is concerned with cluster consensus control of multiagent systems (MASs) with the fixed/switching topology under a dynamic event-trigger (DET) mechanism. A refined sampled-data-based DET scheme is proposed by introducing two dynamically adjusting threshold parameters to distinguish the different transmission requirements for neighboring agents intra and outer cluster. Faced with the difficulties of acquiring full state information among spatially distributed agents, output feedback is employed to construct cooperative control protocols. Both fixed and switching topologies are considered to execute the designed DET-based cooperative cluster consensus control protocols. By constructing appropriate Lyapunov–Krasovskii functionals (LKFs), some sufficient criteria in terms of matrix inequalities for the cluster consensus of MASs are derived, which can ensure that the error system with the proposed DET-based control strategy is asymptotically stable. Facing the nonconvex issue induced by output feedback, a particle swarm optimization (PSO)-based control design algorithm is novelly developed to calculate the control gains and event-triggering parameters jointly based on the derived stability criteria. The elements of the matrix variables are valued stochastically in certain ranges and the fitness function is designed as the accumulation of the weighting value of each matrix inequality. Finally, an application of multiple satellites formation flying is applied to numerically illustrate the effectiveness of the cluster consensus control strategy with the designed DET mechanism.

This article is concerned with quasi-synchronization of grid-connected systems in electrical networks, where heterogeneous Inductance-Capacitance (LC) oscillators are coupled via electrical inductance subject to time-varying delays. Note that complete synchronization fails to be accomplished due to the existence of nonidentical parameters and quasi-synchronization cannot be achieved via non-delayed inductive coupling. A configuration of multiple heterogeneous LC oscillators with inductive coupling subject to time-varying lumped delay is first constructed by introducing an active delay intentionally. Then the complete-type Lyapunov-Krasovskii functionals (LKF) are constructed to investigate the exponential convergence of quasi-synchronization of LC oscillators in the presence of parameter mismatches utilizing the positive effects of interval time-varying delays. Some feasible synchronization criteria are derived. The gain matrix can be designed by solving a set of linear matrix inequalities combining an optimization algorithm. Finally, a numerical example of five LC oscillators in photovoltaic grid-connected system is given to demonstrate the effectiveness of the proposed method.