Spring Semester of the 2019-2020 Academic Year
The tentative schedule for the spring semester of 2019-2020 is given as follows:
- Peng JING
- Lepeng MA
- Jiawei ZHU
- Wenbo HU
- Chengwang YANG
- Xiao WANG
- Chunxiang JIA
- Jin JIN
- Qingfu CUI
- Wuguang WANG
- Yanying YU
- Shuocheng KANG
- Cunhao WEI
- Speaker: Chunxiang Jia
- Title: Distributed Optimal Formation Control with Hard Constraints on Energy and Time
- Abstract: This paper studies distributed optimal formation control with hard constraints on energy levels and termination time, in which the formation error is to be minimized jointly with the energy cost. The main contributions include a globally optimal distributed formation control law and a thorough analysis of the roles of energy, time and control in limiting formation control feasibility. These results characterize and quantify explicitly how energy levels, task termination time, steady-state error tolerance, as well as the network topology may impose inherent limitations on the feasibility of the formation control mission. Most notably, the lower bounds on the termination time and the minimum energy levels are derived, which are given in terms of the initial formation error, the parameter of steady-state error tolerance, and the largest eigenvalue of the Laplacian matrix. These lower bounds can be employed to assert whether a energy and time constrained formation control task is achievable and how to accomplish such a task.
- Speaker: Xiao Wang
- Title: Fragility Limits Performance in Complex Networks
- Abstract: While numerous studies have suggested that large natural, biological, social, and technological networks are fragile, convincing theories are still lacking to explain why natural evolution and human design have failed to optimize networks and avoid fragility. In this paper we provide analytical and numerical evidence that a tradeoff exists in networks with linear dynamics, according to which general measures of robustness and performance are in fact competitive features that cannot be simultaneously optimized. Our findings show that large networks can either be robust to variations of their weights and parameters, or efficient in responding to external stimuli, processing noise, or transmitting information across long distances. As illustrated in our numerical studies, this performance tradeoff seems agnostic to the specific application domain, and in fact it applies to simplified models of ecological, neuronal, and traffic networks.
- Speaker: Chengwang Yang
- Title: Distributed Map Merging with Consensus on Common Information
- Abstract: Sensor fusion methods combine noisy measurements of common variables observed by several sensors, typically by averaging information matrices and vectors of the measurements. Some sensors may have also observed exclusive variables on their own. Examples include robots exploring different areas or cameras observing different parts of the scene in map merging or multi-target tracking scenarios. Iteratively averaging exclusive information is not efficient, since only one sensor provides the data, and the remaining ones echo this information. This paper proposes a method to average the information matrices and vectors associated only to the common variables. Sensors use this averaged common information to locally estimate the exclusive variables. Our estimates are equivalent to the ones obtained by averaging the complete information matrices and vectors. The proposed method preserves properties of convergence, unbiased mean, and consistency, and improves the memory, communication, and computation costs.
- Speaker: Wenbo Hu
- Title: Multi-Agent Reinforcement Learning For Networked System Control
- Abstract: This paper considers multi-agent reinforcement learning (MARL) in networked system control. Specifically, each agent learns a decentralized control policy based on local observations and messages from connected neighbors. We formulate such a networked MARL (NMARL) problem as a spatiotemporal Markov decision process and introduce a spatial discount factor to stabilize the training of each local agent. Further, we propose a new differentiable communication protocol, called NeurComm, to reduce information loss and non-stationarity in NMARL. Based on experiments in realistic NMARL scenarios of adaptive traffic signal control and cooperative adaptive cruise control, an appropriate spatial discount factor effectively enhances the learning curves of non-communicative MARL algorithms, while NeurComm outperforms existing communication protocols in both learning efficiency and control performance.
- Speaker: Jiawei Zhu
- Title: Knowledge-Based Prediction of Network Controllability Robustness
- Abstract: Network controllability robustness reflects how well a networked system can maintain its controllability against destructive attacks. Its measure is quantified by a sequence of values that record the remaining controllability of the network after a sequence of node-removal or edge-removal attacks. Traditionally, the controllability robustness is studied only for directed networks and is determined by attack simulations, which is computationally time consuming or even infeasible. In the present paper, an improved method for predicting the controllability robustness of undirected networks is developed based on machine learning using a group of convolutional neural networks (CNNs). In this scheme, a number of training data generated by simulations are used to train the group of CNNs for classification and prediction, respectively. Extensive experimental studies are carried out, which demonstrate that 1) the proposed method predicts more precisely than the classical single-CNN predictor; 2) the proposed CNN-based predictor provides a better predictive measure than the traditional spectral measures and network heterogeneity.
- Speaker: Lepeng Ma
- Title: Fully Convolutional Networks for Semantic Segmentation
- Abstract: Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixelstopixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build "fully convolutional" networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet , the VGG net , and GoogLeNet ) into fully convolutional networks and transfer their learned representations by fine-tuning  to the segmentation task. We then define a skip architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves stateof-the-art segmentation of PASCAL VOC (20% relative improvement to 62.2% mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes less than one fifth of a second for a typical image.
- Speaker: Peng Jing
- Title: Optimal Stationary Synchronization of Heterogeneous Linear Multi-Agent Systems
- Abstract: In this paper, we address the output synchronization of heterogeneous linear networks. In the literature, all agents are typically required to synchronize exactly to a common trajectory. Here, we introduce optimal stationary synchronization (OSS) instead which permits non-zero steady state synchronization errors. As a benefit, we are able to relax standard requirements. E.g., agents are allowed to participate in the network even when they usually cannot synchronize exactly. In addition, OSS enables agents to save input-energy by synchronizing within tolerable error-bounds. Our new method combines the synchronization of bounded exosystems with local infinite-time linear quadratic tracking (LQT). This results in an optimal balance of each agent's synchronization error versus its consumed input-energy. Moreover, we extend recent results in LQT such that the derived time-invariant optimal control guarantees that the synchronization error satisfies given strict bounds. All these aspects are demonstrated by an illustrative simulation example with a detailed analysis.
- Speaker: Cunhao Wei
- Title: Leveraging Diversity for Achieving Resilient Consensus in Sparse Networks
- Abstract: A networked system can be made resilient against adversaries and attacks if the underlying network graph is structurally robust. For instance, to achieve distributed consensus in the presence of adversaries, the underlying network graph needs to satisfy certain robustness conditions. A typical approach to making networks structurally robust is to strategically add extra links between nodes, which might be prohibitively expensive. In this paper, we propose an alternative way of improving networks robustness, that is by considering heterogeneity of nodes. Nodes in a network can be of different types and can have multiple variants. As a result, different nodes can have disjoint sets of vulnerabilities, which means that an attacker can only compromise a particular type of nodes by exploiting a particular vulnerability. We show that, by such a diversification of nodes, attackers ability to change the underlying network structure is significantly reduced. Consequently, even a sparse network with heterogeneous nodes can exhibit the properties of a structurally robust network. Using these ideas, we propose a distributed control policy that utilizes heterogeneity in the network to achieve resilient consensus in adversarial environment. We extend the notion of (r; s)-robustness to incorporate the diversity of nodes and provide necessary and sufficient conditions to guarantee resilient distributed consensus in heterogeneous networks. Finally we study the properties and construction of robust graphs with heterogeneous nodes.
- Speaker: ShuoCheng Kang
- Title: Distributed optimal control for multiple high-speed train movement An alternating direction method of multipliers
- Abstract: This study systematically investigates the distributed optimal control of multiple high-speed train movement to overcome communication constraints and realize efficient speed control. A dynamic multiple train movement model is constructed to capture the dynamic evolution of trains in real-world operations. With the coupling constraints for the safe distance headway among neighboring trains, an optimal control problem is formulated to improve speed and distance tracking accuracy for each train and reduce energy consumption. Based on the dual decomposition technique, each train is regarded as a subsystem equipped with a local controller while a centralized entity manages the subsystem via the prices associated with the coupling constraints. Within the framework of an alternating direction method of multipliers and a model predictive control method, a novel distributed optimal control algorithm based on a distributed message passing mechanism is designed. The algorithm divides the original complex optimal control problem into many smaller optimal control problems that can be computed in parallel and satisfies the real-time control requirement. Numerical examples are provided to illustrate the effectiveness of the proposed train control model and methods.
- Speaker: Yanying Yu
- Title: Towards Optimal Robustness of Network Controllability: An Empirical Necessary Condition
- Abstract: To better understand the correlation between network topological features and the robustness of network controllability in a general setting, this paper suggests a practical approach to searching for optimal network topologies with given numbers of nodes and edges. Since theoretical analysis seems impossible at least in the present time, exhaustive search based on optimization techniques is employed, firstly for a group of smallsized networks that are realistically workable, where exhaustive means 1) all possible network structures with the given numbers of nodes and edges are computed and compared, and 2) all possible node-removal sequences are considered. A main contribution of this paper is the observation of an empirical necessary condition (ENC) from the results of exhaustive search, which shrinks the search space to quickly find an optimal solution. ENC shows that the maximum and minimum in- and out-degrees of an optimal network structure should be almost identical, or within a very narrow range, i.e., the network should be extremely homogeneous. Edge rectification towards the satisfaction of the ENC is then designed and evaluated. Simulation results on largesized synthetic and real-world networks verify the effectiveness of both the observed ENC and the edge rectification scheme. As more operations of edge rectification are performed, the network is getting closer to exactly satisfying the ENC, and consequently the robustness of the network controllability is enhanced towards optimum.
- Speaker: Wuguang Wang
- Title: Robust Dynamic Average Consensus with Prescribed Performance
- Abstract: We consider the dynamic average consensus problem in which multiple agents cooperate in order to track the average of locally available time-varying reference signals. Within this framework, each agent is only capable of local computations and local communication with its neighbors. Hence, we propose a distributed estimation procedure that guarantees bounded tracking error, with practical asymptotic convergence at the steady state even for fast time-varying reference signals. The transient needed for the agents to reach consensus and the maximum deviation among their estimates can be set arbitrarily small a priori, via the appropriate selection of certain design parameters. Moreover, the consensus and the tracking performance are regulated independently. Finally, we validate the proposed scheme through various simulated paradigms.
- Title: Robust Event-triggered Dynamic Average Consensus against Communication Link Failures with Application to Battery Control
- Abstract: Dynamic average consensus (DAC) has found applications in various systems. The existing event-triggered DAC algorithms have not well addressed the issue of key communication link failures that lead to the separation of the initial communication topology. This paper presents a modified eventtriggered DAC algorithm which is independent of its initial conditions. As a result, it is robust against key communication link failures. In this algorithm, each agent decides locally when to transmit signals to its neighbours. In this way, the communication burden among the neighboring agents is reduced. A numerical example is provided to illustrate the effectiveness of the proposed algorithm. Moreover, the proposed algorithm is applied to a stateof- charge balance control problem of batteries in energy systems, and both simulations and hardware in the loop (HIL) tests are provided to demonstrate the control performance.
- Speaker: Yong Du
- Title: Active Localization of Gas Leaks Using Fluid Simulation
- Abstract: Sensors are routinely mounted on robots to acquire various forms of measurements in spatiotemporal fields. Locating features within these fields and reconstruction (mapping) of the dense fields can be challenging in resource-constrained situations,such as when trying to locate the source of a gas leak from a small number of measurements. In such cases, a model of the underlying complex dynamics can be exploited to discover informative paths within the field. We use a fluid simulator as a model to guide inference for the location of a gas leak. We perform localization via minimization of the discrepancy between observed measurements and gas concentrations predicted by the simulator. Our method is able to account for dynamically varying parameters of wind flow(e.g., direction and strength) and its effects on the observed distribution of gas. We develop algorithms for offline inference as well as for online path discovery via active sensing. We demonstrate the efficiency, accuracy, and versatility of our algorithm using experiments with a physical robot conducted in outdoor environments.We deploy an unmanned air vehicle mounted with a CO2 sensor to automatically seek out a gas cylinder emitting CO2 via a nozzle. We evaluate the accuracy of our algorithm by measuring the error in the inferred location of the nozzle, based on which we show that our proposed approach is competitive with respect to state-of-the-art baselines.
- Speaker: Qingfu Cui
- Title: Optimal Delay Control for Combating Bufferbloat in the Internet
- Abstract: Nowadays, inexpensive memory has led to large router buffers in the Internet, leading to excessively long queues and high latency. This phenomenon is recently termed as "bufferbloat". Recently, Controlled Delay (CoDel) has been proposed as an active queue management (AQM) algorithm to address this problem by setting a limit (i.e., a target delay) on the queueing delay. However, such a target delay is fixed, and CoDel is thus unable to adapt to changes in the network environment. In this paper, we propose a framework to adaptively control the target delay with changing network conditions, so as to overcome bufferbloat. Specifically, we first develop a theoretical model to analyze the stability of CoDel and derive the stable range of the target delay using control theory. We then propose Adaptive CoDel, which dynamically selects an optimal value of the target delay in the stable range, such that the aggregate TCP goodput is maximized, while keeping the system stable. Our simulation results show that Adaptive CoDel not only effectively controls the queueing delay, but also achieves higher aggregate TCP goodput compared to other AQM algorithms.
- Speaker: Jin Jin
- Title: A Randomized Block Coordinate Iterative Regularized Subgradient Method for High-dimensional Ill-posed Convex Optimization
- Abstract: Motivated by ill-posed optimization problems arising in image processing, we consider a bilevel optimization model, where we seek among the optimal solutions of the inner level problem, a solution that minimizes a secondary metric. Minimal norm gradient, sequential averaging, and iterative regularization appear among the known schemes developed for addressing this class of problems. However, to the best of our knowledge, none of these schemes address nondifferentiability and high-dimensionality of the solution space. Motivated by this gap, we consider the case where the solution space has a block structure and both objective functions are nondifferentiable. We develop a randomized block coordinate iterative regularized subgradient scheme (RB-IRG). Under a uniform distribution for selecting the blocks and a careful choice of the stepsize and regularization sequences, we establish the convergence of the sequence generated by RB-IRG scheme to the unique solution of the bilevel problem of interest in an almost sure sense. Furthermore, we derive a convergence rate in terms of the expected objective value of the inner level problem, where d denotes the number of blocks and \delta > 0 is an arbitrary small scalar. We demonstrate the performance of RB-IRG algorithm in solving the ill-posed problems arising in image processing.
- Speaker: Chunxiang Jia
- Title: Active-Passive Dynamic Consensus Filters for Linear Time-Invariant Multiagent Systems
- Abstract: Active-passive dynamic consensus filters consist of a group of agents, where a subset of these agents are able to observe a quantity of interest (i.e. active agents) and the rest are subject to no observations (i.e. passive agents). Specifically, the objective of these filters is that the states of all agents are required to converge to the weighted average of the set of observations sensed by the active agents. Existing active-passive dynamic consensus filters in the classical sense assume that all agents can be modeled as having single integrator dynamics, which may not always hold in practice. Motivating from this standpoint, the contribution of this paper is to introduce a new class of active-passive dynamic consensus filters, where agents have (homogeneous) linear time-invariant dynamics. We demonstrate that for output controllable agents, the output of all active and passive agents converge to a neighborhood of the weighted average of the set of applied exogenous inputs. A numerical example is also given to illustrate the efficacy of the presented theoretical results.
- Speaker: Xiao Wang
- Title: Universal resilience patterns in complex networks
- Abstract: Resilience, a system's ability to adjust its activity to retain its basic functionality when errors, failures and environmental changes occur, is a defining property of many complex systems1. Despite widespread consequences for human health2, the economy3 and the environment4, events leading to loss of resilience-from cascading failures in technological systems5 to mass extinctions in ecological networks6-are rarely predictable and are often irreversible. These limitations are rooted in a theoretical gap: the current analytical framework of resilience is designed to treat low-dimensional models with a few interacting components7, and is unsuitable for multi-dimensional systems consisting of a large number of components that interact through a complex network. Here we bridge this theoretical gap by developing a set of analytical tools with which to identify the natural control and state parameters of a multi-dimensional complex system, helping us derive effective one-dimensional dynamics that accurately predict the system's resilience. The proposed analytical framework allows us systematically to separate the roles of the system's dynamics and topology, collapsing the behaviour of different networks onto a single universal resilience function. The analytical results unveil the network characteristics that can enhance or diminish resilience, offering ways to prevent the collapse of ecological, biological or economic systems, and guiding the design of technological systems resilient to both internal failures and environmental changes.
- Speaker: Chengwang Yang
- Title: An AEKF-SLAM Algorithm with Recursive Noise Statistic Based on MLE and EM
- Abstract: Extended Kalman Filter (EKF) has been popularly utilized for solving Simultaneous Localization and Mapping (SLAM) problem. Essentially, it requires the accurate system model and known noise statistic. Nevertheless, this condition can be satisfied in simulation case. Hence, EKF has to be enhanced when it is applied in the real-application. Mainly, this improvement is known as adaptive-based approach. In many different cases, it is indicated by some manners of estimating for either part or full noise statistic. This paper present a proposed method based on the adaptive-based solution used for improving classical EKF namely An Adaptive Extended Kalman Filter. Initially, the classical EKF was improved based on Maximum Likelihood Estimation (MLE) and Expectation-Maximization (EM) Creation. It aims to equips the conventional EKF with ability of approximating noise statistic and its covariance matrices recursively. Moreover, EKF was modified and improved to tune the estimated values given by MLE and EM creation. Besides that, the recursive noise statistic estimators were also estimated based on the unbiased estimation. Although it results high quality solution but it is followed with some risks of non-positive definite matrices of the process and measurement noise statistic covariances. Thus, an addition of Innovation Covariance Estimation (ICE) was also utilized to depress this possibilities. The proposed method is applied for solving SLAM problem of autonomous wheeled mobile robot. Henceforth, it is termed as AEKF-SLAM Algorithm. In order to validate the effectiveness of proposed method, some different SLAM-Based algorithm were compared and analyzed. The different simulation has been showing that the proposed method has better stability and accuracy compared to the conventional filter in term of Root Mean Square Error (RMSE) of Estimated Map Coordinate (EMC) and Estimated Path Coordinate (EPC).