查看更多>>摘要：The overall performance of multi-robot collaborative systems is significantly affected by the multi-robot task allocation.To improve the effectiveness,robustness,and safety of multi-robot collaborative systems,a multimodal multi-objective evolutionary algorithm based on deep reinforcement learning is proposed in this paper.The improved multimodal multi-objective evolutionary algorithm is used to solve multi-robot task allo-cation problems.Moreover,a deep reinforcement learning strategy is used in the last generation to provide a high-quality path for each assigned robot via an end-to-end manner.Comparisons with three popular multimodal multi-objective evolutionary algorithms on three different scenarios of multi-robot task allocation problems are carried out to verify the performance of the proposed algorithm.The experimental test results show that the proposed algorithm can generate sufficient equivalent schemes to improve the availability and robustness of multi-robot collaborative systems in uncertain environments,and also produce the best scheme to improve the overall task execution efficiency of multi-robot collaborative systems.
查看更多>>摘要：In order to solve the problem of small object size and low detection accuracy under the unmanned aerial vehicle(UAV)platform,the object detection algorithm based on deep aggregation network and high-resolution fusion module is studied.Furthermore,a joint network of object detection and feature extraction is studied to construct a real-time multi-object tracking algorithm.For the problem of object association failure caused by UAV movement,image registration is applied to multi-object tracking and a camera motion discrimination model is proposed to improve the speed of the multi-object tracking algorithm.The simulation results show that the algorithm proposed in this study can improve the accuracy of multi-object tracking under the UAV platform,and effectively solve the problem of association failure caused by UAV movement.
查看更多>>摘要：Visual object tracking is an important issue that has received long-term attention in computer vision.The ability to effectively handle occlusion,especially severe occlusion,is an important aspect of evaluating the performance of object tracking algorithms in long-term tracking,and is of great significance to improving the robustness of object tracking algorithms.However,most object tracking algorithms lack a processing mechanism specifically for occlusion.In the case of occlusion,due to the lack of target information,it is necessary to predict the target position based on the motion trajectory.Kalman filtering and particle filtering can effectively predict the target motion state based on the historical motion information.A single object tracking method,called probabilistic discriminative model prediction(PrDiMP),is based on the spatial attention mechanism in complex scenes and occlusions.In order to improve the performance of PrDiMP,Kalman filtering,particle filtering and linear filtering are introduced.First,for the occlusion situation,Kalman filtering and particle filtering are respectively introduced to predict the object position,thereby replacing the detection result of the original tracking algorithm and stopping recursion of target model.Second,for detection-jump problem of similar objects in complex scenes,a linear filtering window is added.The evaluation results on the three datasets,including GOT-10k,UAV123 and LaSOT,and the visualization results on several videos,show that our algorithms have improved tracking performance under occlusion and the detection-jump is effectively suppressed.
查看更多>>摘要：Infrared detection technology has the advantages of all-weather detection and good concealment,which is widely used in long-distance target detection and tracking systems.However,the complex background,the strong noise,and the characteristics of small scale and weak intensity of targets bring great difficulties to the detection of infrared small targets.A multi-channel based on attention network is proposed in this paper,aimed at the problem of high missed detection rate and false alarm rate of traditional algorithms and the problem of large model,high complexity and poor detection performance of deep learning algorithms.First,given the difficulty in extracting the features of infrared multiscale and small dim targets,the multiple channels are designed based on dilated convolution to capture multiscale target features.Second,the coordinate attention block is incorporated in each channel to suppress background clutters adaptively and enhance target features.In addition,the fusion of shallow detail features and deep abstract semantic features is realized by synthesizing the contextual attention fusion block.Finally,it is verified that,compared with other state-of-the-art methods based on the datasets SIRST and MDFA,the proposed algorithm further improves the detection effect,and the model size and computational complexity are smaller.
查看更多>>摘要：Mobile robot local motion planning is responsible for the fast and smooth obstacle avoidance,which is one of the main indicators for evaluating mobile robots'navigation capabilities.Current methods formulate local motion planning as a unified problem;therefore it cannot satisfy the real-time requirement on the platform with limited computing ability.In order to solve this problem,this paper proposes a fast local motion planning method that can reach a planning frequency of 500 Hz on a low-cost CPU.The proposed method decouples the local motion planning as the front-end path searching and the back-end optimization.The front-end is composed of the environment topology analysis and graph searching.The back-end includes dynamically feasible trajectory generation and optimal trajectory selection.Different from the popular methods,the proposed method decomposes the local motion planning into four sub-modules,each of which aims to solve one problem.Combining four sub-modules,the proposed method can obtain the complete local motion planning algorithm which can fast generate a smooth and collision-free trajectory.The experimental results demonstrate that the proposed method has the ability to obtain the smooth,dynamically feasible and collision-free trajectory and the speed of the planning is fast.
查看更多>>摘要：For a class of discrete-time singular stochastic systems with multi-state delay,the stabilization problem of receding horizon control(RHC)is concerned.Due to the difficulty in solving the proposed optimization problem,the RHC stabilization for such systems has not been solved.By adopting the forward and backward equation technique,the optimization problem is solved completely.A sufficient and necessary condition for the optimization controller to have a unique solution is given when the regularization and pulse-free conditions are satisfied.Based on this controller,an RHC stabilization condition is derived,which is in the form of linear matrix inequality.It is proved that the singular stochastic system with multi-state delay is stable in the mean-square sense under appropriate assumptions when the terminal weighting matrix satisfies the given inequality.Numerical examples show that the proposed RHC method is effective in stabilizing singular stochastic systems with multi-state delay.
查看更多>>摘要：Due to the diversity of work requirements and environment,the number of degrees of freedom(DOFs)and the complexity of structure of industrial robots are constantly increasing.It is difficult to establish the accurate dynamical model of industrial robots,which greatly hinders the realization of a stable,fast and accurate trajectory tracking control.Therefore,the general expression of the constraint relation in the explicit dynamic equation of the multi-DOF industrial robot is derived on the basis of solving the Jacobian matrix and Hessian matrix by using the kinematic influence coefficients method.Moreover,an explicit dynamic equation with general constraint relation expression is established based on the Udwadia-Kalaba theory.The problem of increasing the time of establishing constraint relationship when the multi-DOF industrial robots complete complex task constraints is solved.With the SCARA robot as the research object,the simulation results show that the proposed method can provide a new idea for industrial robot system modeling with complex constraints.
查看更多>>摘要：Beam position monitors(BPMs)are important to monitor the beam moving steadily.Keeping the beam's normal motion is an important mission for Shanghai Synchrotron Radiation Facility.Effective diagnostic analysis is an important way to accomplish this task.This paper develops a new method based on clustering analysis to diagnose the healthy of BPMs with basic running data,i.e.,the β oscillation of X and Y directions and noise data.The analysis results showed that all beam position monitors(140 BPMs)can be classified into three groups:normal group,worse performance group,and fault group,respectively.In addition,the abnormal BPMs(including worse performance)could be marked.The new method showed its ability to handle faulty BPMs and it could instruct daily maintenance.On the other hand,it will be a useful supplement for data analysis in accelerator physics.
查看更多>>摘要：Traveling salesman problem(TSP)is a classic non-deterministic polynomial-hard optimization prob-lem.Based on the characteristics of self-organizing mapping(SOM)network,this paper proposes an improved SOM network from the perspectives of network update strategy,initialization method,and parameter selection.This paper compares the performance of the proposed algorithms with the performance of existing SOM network algorithms on the TSP and compares them with several heuristic algorithms.Simulations show that compared with existing SOM networks,the improved SOM network proposed in this paper improves the convergence rate and algorithm accuracy.Compared with iterated local search and heuristic algorithms,the improved SOM net-work algorithms proposed in this paper have the advantage of fast calculation speed on medium-scale TSP.
查看更多>>摘要：Based on option-critic algorithm,a new adversarial algorithm named deterministic policy network with option architecture is proposed to improve agent's performance against opponent with fixed offensive algorithm.An option network is introduced in upper level design,which can generate activated signal from defensive and of-fensive strategies according to temporary situation.Then the lower level executive layer can figure out interactive action with guidance of activated signal,and the value of both activated signal and interactive action is evaluated by critic structure together.This method could release requirement of semi Markov decision process effectively and eventually simplified network structure by eliminating termination possibility layer.According to the result of experiment,it is proved that new algorithm switches strategy style between offensive and defensive ones neatly and acquires more reward from environment than classical deep deterministic policy gradient algorithm does.
NETL
NSTL
万方数据