Xisheng Feng

Constrained Multi-objective Task Assignment for UUVs Using Multiple Ant Colonies System

The purpose of this research is to develop an effective task assignment algorithm for multiple unmanned underwater vehicles (UUVs) to reacquaint multiple targets. This algorithm is specifically designed for the underwater environment where vehicles typically have dissimilar starting and ending locations. Besides the objective of minimizing the total distance of multiple vehicles, the objectives of minimizing the total turning angle and the constraint of balancing the targets number visited by each vehicle are also considered. This problem is modeled as a constrained multi-objective MTSP. The different measurement units and order of magnitudes of multiple objectives significantly increase the difficulty to generate an effective solution. The proposed algorithm consists of two phases: task number assignment and task assignment using multiple ant colony system (MACS) which is extended from the classical ant colony system (ACS). In the first phase, the target number is assigned to each vehicle. Afterwards, MACS is used to solve constrained multi-objective MTSP, in which multiple ant colonies work separately to optimize dissimilar objectives, the ideal solution is generated according to the result of each colony, and the output is the best solution which has the smallest deviation from the ideal solution in the set of Pareto optimal solutions. The computational results show that the output of the proposed algorithm can satisfy the constrained multi-objective requirement and can be applied to underwater application scenario.

Autonomic element based architecture for unmanned underwater vehicles

The architecture has always been one of the key aspects for designing an unmanned underwater vehicle. The architecture should serve as an aid, not a burden, in the integration of modules that have been developed independently, so it must not be overly restrictive. In this article, three types of architectures (deliberative, reactive, and hybrid architecture) are reviewed. Then the criteria for evaluating architectures are discussed. By borrowing the idea of autonomic computing, the autonomic element based architecture for unmanned underwater vehicles is constructed. Finally, simulations are carried out on a semi-physical platform to validate the feasibility of this architecture.

Reliable transmission scheduling for multi-channel wireless sensor networks with low-cost channel estimation

The shared-medium nature and complex wireless environment of wireless sensor networks (WSNs) poses fundamental challenges to the design of effective transmission scheduling algorithms that are optimised with respect to superframe length and reliability. In this study, the authors propose an adaptive and reliable transmission scheduling algorithm for WSNs based on low-cost estimation of channel states. The authors establish a hierarchical scheduling framework on global centralised timeslot scheduling and local distributed channel scheduling. On the one hand, global centralised timeslot scheduling aims to guarantee global optimality of resource allocation, during which a mathematical reliability model is built to avoid resource waste by the stationary allocation method and improve the reliability of packet transmission. On the other hand, local distributed channel scheduling shares the responsibility of resource allocation. During channel scheduling, the channel model is constructed by the dynamic programming method and takes both probing cost and channel quality into consideration, which alleviates the uncertain and time-varying interference and overcomes the blindness of traditional methods. In contrast with previous works that do not consider link reliability and channel probing cost and often assume two channel states, the scheduling algorithm performs reliably for an arbitrary number of channels and arbitrary number of channel states. Extensive simulations and experiments under a variety of network environments have been conducted to validate our theoretical claims.

The development of autonomous underwater vehicle's semi-physical virtual reality system

A semi-physical virtual reality system applying to autonomous underwater vehicle control study is presented in this paper. This system is composed of three nodes: (1) autopilot node for mission planning and control, (2) virtual node for virtual devices and sensors and (3) visualization node for virtual ocean environment and display the state of AUV. The autopilot node can be both linked to the real sensors and actuators dealing with a real mission, and linked to the virtual world for simulation. The advantages of this semi-physical virtual reality system include that: (1) the vehicle software can he simulated in the laboratory before real experiments taking place, (2) the merit and demerit of some newly AUV system can be confirmed, (3) Moreover intensive testing is easy and its cost is not too expensive.

Local Autoencoding for Parameter Estimation in a Hidden Potts-Markov Random Field

A local-autoencoding (LAE) method is proposed for the parameter estimation in a Hidden Potts-Markov random field model. Due to sampling cost, Markov chain Monte Carlo methods are rarely used in real-time applications. Like other heuristic methods, LAE is based on a conditional independence assumption. It adapts, however, the parameters in a block-by-block style with a simple Hebbian learning rule. Experiments with given label fields show that the LAE is able to converge in far less time than required for a scan. It is also possible to derive an estimate for LAE based on a Cramer-Rao bound that is similar to the classical maximum pseudolikelihood method. As a general algorithm, LAE can be used to estimate the parameters in anisotropic label fields. Furthermore, LAE is not limited to the classical Potts model and can be applied to other types of Potts models by simple label field transformations and straightforward learning rule extensions. Experimental results on image segmentations demonstrate the efficiency and generality of the LAE algorithm.

Optimal Convergecast Scheduling Limits for Clustered Industrial Wireless Sensor Networks

Increased mobility coupled with a possible reduction of cabling costs and deployment time makes wireless communication an attractive alternative for the industrial process monitoring and control. The major obstacles toward the utilization of wireless industrial networks are predominantly the timeliness and reliability requirements. In this paper, orienting to clustered industrial wireless sensor networks, we analyze the performance bounds of the convergecast scheduling, which is a typical many-to-one communication in wireless industrial networks. The analysis aims to the cluster-line and cluster-tree topologies. Each kind is future divided into three scenarios according to an application parameter, named data update rate. Firstly, we establish the lower bounds on the number of timeslots to finish the intracluster and the intercluster convergecast transmissions. Secondly, we establish the lower bounds on the number of channels based on the lower bounds on the number of timeslots and maximum available channels in a multichannel scenario. Lastly, we carry out the extensive analysis-taking packet retransmissions into consideration so as to meet the reliability requirement. Experiment results validate the correctness and tightness of our theory analysis.

Toward a generalized architecture for unmanned underwater vehicles

A common feature of unmanned vehicles is their complexity, which grows apace and provides its own challenges. Frameworks for managing this growing complexity have always been one of the key aspects of designing an unmanned vehicle. In this paper, a generalized architecture is proposed to not only address the complexity in developing an unmanned vehicle, but also support the algorithm exchange and technology transfer for integrating efforts from different researchers. We first detail the autonomic element, which is the fundamental unit of the architecture. Then the architecture is constructed, and thorough discussions are given. Finally, simulations on a semi-physical platform are carried out to examine the performance of this architecture.

The Design and Development of Simulator System—For Manned Submersible Vehicle

A semi-physical virtual reality system applying to MSV (manned submersible vehicle) control study is presented in this paper. This system is composed of three nodes: (1) control node for receiving sensors information and sending control command, (2) virtual sensors node for virtual devices and sensors and (3) Visualization node for virtual ocean environment and displaying the posture of MSV. The control node can be both linked to the real sensors and actuators dealing with a real mission, and linked to the virtual world for test purpose. The advantages of this semi-physical virtual reality system include that: (1) the MSV software can be tested in the laboratory before real experiments taking place, (2) the merit and demerit of some newly underwater vehicle system can be verified, (3) Moreover intensive testing is easy and its cost is not too expensive

Label field initialization for MRF-based sonar image segmentation by selective autoencoding

The optimal solution of a Markov random field (MRF) can be solved by constructing a Markov chain that eventually goes to a balance state. However, in most situations, only an suboptimal solution can be obtained, because it is hard to choose the ideal initial state and the updating strategy. While the updating strategy has been extensively investigated, the initialization issue has been fully neglected. Though k-means-clustering has been used exclusively in initializing the label field, it suffers from the lack of account of the local constraints, which is the most essential part of the MRF model. A structural method based on selective autoencoding (SAE) is proposed for the label field initialization of MRF model in the task of sonar image segmentation. SAE is similar to the AutoEncoder, with the largest difference on the activation function, where a piece-wise sigmoid activation function with two different slop parameters is used to selectively encode image patches that resemble shadow ares or other areas. The synapse matrixes of SAE network act as information filters, preserve specific area adaptively and selectively, generating a label field that is much closer to the balance state. Experiments on sonar image segmentation demonstrate the efficiency of the SAE algorithm.

Research on the decentralized supervisory control of autonomous underwater vehicle

In order to independently complete some situation-adapted missions in unknown undersea environment, higher-level decision techniques are needed to provide an autonomous underwater vehicle (AUV) with the ability of active autonomy. This paper uses certain aspects of the RW supervisory control theory to investigate the decentralized supervisory control (DSC) of AUV for the basal achievement of active autonomy. The logical hierarchical model of AUV control system in terms of finite state automata is built, and the realization of DSC is given in detail. The experiment results such as on-line global and local path planning on the semi-physical simulation platform of AUV are illustrated at the end.