Search

In this paper, a data-driven control approach is developed by reinforcement learning &#x0028 RL &#x0029 to solve the global robust optimal output regulation problem &#x0028 GROORP &#x0029 of partially linear systems with both static uncertainties and nonlinear dynamic uncertainties. By developing a proper feedforward controller, the GROORP is converted into a global robust optimal stabilization problem. A robust optimal feedback controller is designed which is able to stabilize the system in the presence of dynamic uncertainties. The closed-loop system is assured to be input-to-output stable regarding the static uncertainty as the external input. This robust optimal controller is numerically approximated via RL. Nonlinear small-gain theory is applied to show the input-to-output stability for the closed-loop system and thus solves the original GROORP. Simulation results validates the efficacy of the proposed methodology.

This paper presents a cooperative/shared framework of the driver and his/her semi-autonomous vehicle in order to achieve desired steering performance. In particular, a copilot controller and the driver together operate and control the vehicle. Exploiting the classical small-gain theory, our proposed shared steering controller is developed independent of the unmeasurable internal states of the human driver, and only relies on his/her steering torque. Furthermore, by adopting data-driven adaptive dynamic programming and an iterative learning scheme, the shared steering controller is studied from the measurable data of the driver and the vehicle. Meanwhile, the accurate knowledge of the driver and the vehicle dynamics is unnecessary, which settles the problem of their potential parametric variations/uncertainties in practice. The effectiveness of the proposed method is validated by rigorous analysis and demonstrated by numerical simulations.

This paper investigates the consensus problem of general linear discrete-time multiagent systems by using distributed model predictive control (DMPC) with self-triggered mechanism. First, a novel DMPC-based consensus algorithm is proposed, where each agent only needs to obtain its neighbors’ predicted state sequences once at each time step. We prove that the resultant DMPC optimization problem is feasible, and the proposed algorithm guarantees the dynamic consensus of agents. Then, to further reduce the communication cost and the energy consumption of control updates, a self-triggered DMPC-based consensus algorithm is proposed with the control input and the triggering interval jointly optimized. Numerical examples including the benchmark problem with platooning vehicles are provided to verify the effectiveness and advantages of the proposed algorithms.

Widely recognized as an excellent solution of global warming and oil crisis, electric vehicles (EVs), however, suffer remarkable weakness, such as the limited cruise range, which can be partly addressed by introducing en-route charging navigation systems. Different from traditional navigation, which solves a shortest path problem, the en-route charging navigation resorts to a joint charging and routing optimization. In this brief, we formulate the en-route charging navigation in a dynamic programming setting in both a deterministic and a stochastic traffic network. Specifically, to relieve computational complexity in navigation systems, a simplified charge-control (SCC) algorithm is presented in the deterministic case, which can simplify the charging control decisions within an SCC set. In the stochastic case, an online state recursion algorithm is designed, which can provide an accurate navigation utilizing online information. Numerical simulation verifies the computing burden and accuracy of the proposed algorithms in a deterministic and a stochastic network.

Despite the ever-increasing use across different sectors, the lithium-ion batteries (LiBs) have continually seen serious concerns over their thermal vulnerability. The LiB operation involves heat generation and buildup effect, which manifests itself strongly, in the form of highly uneven thermal distribution, for a LiB pack consisting of multiple cells. If not well monitored and managed, the heating may accelerate aging and cause unwanted side reactions. In extreme cases, it will even cause fires and explosions. Toward addressing this threat, this brief, for the first time, seeks to reconstruct the 3-D temperature field of a LiB pack in real time. The major challenge lies in how to acquire a high-fidelity reconstruction with constrained computation time. In this brief, a 3-D thermal model is established first for a LiB pack configured in series, which captures the spatial thermal behavior with a combination of high integrity and low complexity. Given the model, the standard Kalman filter is then distributed to attain temperature field estimation with substantially reduced computational complexity. The arithmetic operation analysis and the numerical simulation illustrate that the proposed distributed estimation achieves a comparable accuracy as the centralized approach but with much less computation.

YAP and TAZ (YAP/TAZ), two major effectors of the Hippo signaling pathway, are frequently activated in human cancers. The activity of YAP/TAZ is strictly repressed upon phosphorylation by LATS1/2 tumor suppressors. However, it is unclear how LATS1/2 are precisely regulated by upstream factors such as Hippo kinases MST1/2. Here, we show that WWC proteins (WWC1/2/3) directly interact with LATS1/2 and SAV1, and SAV1, in turn, brings in MST1/2 to phosphorylate and activate LATS1/2. Hence, WWC1/2/3 play an organizer role in a signaling module that mediates LATS1/2 activation by MST1/2. Moreover, we have defined a minimum protein interaction interface on WWC1/2/3 that is sufficient to activate LATS1/2 in a robust and specific manner. The corresponding minigene, dubbed as SuperHippo, can effectively suppress tumorigenesis in multiple tumor models. Our study has uncovered a molecular mechanism underlying LATS1/2 regulation and provides a strategy for treating diverse malignancies related to Hippo pathway dysregulation.

This paper presents a hierarchical motion planning approach that can provide real-time parking plans for autonomous vehicles with limited memory. Through combining a high-level route planner that searches for collision-free routes given traffic and obstacle information and a low level motion planner that considers vehicle dynamics, our approach generates smooth trajectories with reasonable parking behaviors rapidly with very low memory consumption. This hierarchical approach allows for online path repairing and replanning when newly detected obstacles that were not indicated on the offline map obstruct the original planned trajectory. It employs a fast clearance checking procedure to obtain a practical indicator of repairability as well as heuristic guidance for rapid trajectory repairing, and utilizes the high-level route planner to conduct real-time replanning when trajectory repairing is deemed to be difficult. Performance analysis on parking tasks in simulation environments demonstrates the advantages of the proposed approach in terms of both trajectory quality and planning time.

Mutations in the enzyme isocitrate dehydrogenase 1/2 (IDH1/2) are the most common somatic mutations in low-grade glioma (LGG). The Hippo signaling pathway is known to play a key role in organ size control, and its dysregulation is involved in the development of diverse cancers. Large tumor suppressor 1/2 (LATS1/2) are core Hippo pathway components that phosphorylate and inactivate Yes-associated protein (YAP), a transcriptional co-activator that regulates expression of genes involved in tumorigenesis. A recent report from The Cancer Genome Atlas (TCGA) has highlighted a frequent hypermethylation of LATS2 in IDH-mutant LGG. However, it is unclear if LATS2 hypermethylation is associated with YAP activation and prognosis of LGG patients. Here, we performed a network analysis of the status of the Hippo pathway in IDH-mutant LGG samples and determined its association with cancer prognosis. Combining TCGA data with our biochemical assays, we found hypermethylation of LATS2 promoter in IDH-mutant LGG. LATS2 hypermethylation, however, did not translate into YAP activation but highly correlated with IDH mutation. LATS2 hypermethylation may thus serve as an alternative for IDH mutation in diagnosis and a favorable prognostic factor for LGG patients.

This paper investigates speed-sensorless estimation problem for induction machines, aiming to offer a better balance between speed estimation bandwidth and robustness than a classic adaptive full-order observer (AFO). AFO suffers from a trade-off in selecting its speed adaptation gains: large gains for high bandwidth versus low gains for suppression of ripples induced by model mismatches and noises. We propose two revisions on the AFO to relax the trade-off. First is to adopt a variable speed adaptation gain which is large during transient and is small in steady-state. Second is to include a feedforward term in the speed adaptation law to accommodate the rotor’s mechanical dynamics. An iterative tuning method is presented to adjust feedforward gains, addressing the uncertainties in rotor's inertia and load torque. Experiments show that the proposed method can significantly improve the speed estimation bandwidth while effectively suppressing the fluctuation of the speed estimate during steady state, compared with AFO.

Rotor speed estimation is one of the key problems in speed-sensorless motor drives. Adaptation-based approaches, assuming the rotor speed as a parameter and based on the original coordinates, admit simple estimator designs, albeit suffer from the lack of guaranteed convergence of estimation error dynamics. Focusing on stable speed estimation, this paper proposes a new algorithm based on transforming the motor model into an adaptive observer form via a change of state coordinates. The resultant adaptive estimator renders globally exponentially convergent estimation error dynamics, under persistent excitation condition, which is analogous to entail a time-varying rotor flux. The proposed algorithm is advantageous for its guaranteed stability, ease of tuning, and robustness. Experiments demonstrate its effectiveness in mid-/high-speed operation regions.

We design real-time optimal tracking controllers for servomotor systems engaged in single-axis point-to-point positioning tasks. The design is challenging due to the presence of unmodeled dynamics, along with speed and acceleration constraints. As model-based optimal control design methods cannot be applied directly to this uncertain system, we propose a data-driven approximate dynamic programming approach to learn an optimal tracking controller that is constraint-enforcing. The potential of our proposed method is illustrated on a servomotor that positions the head of a laser drilling machine.

Proteasome inhibitors (PIs) are currently used in the clinic to treat cancers such as multiple myeloma (MM). However, cancer cells often rapidly develop drug resistance towards PIs due to a compensatory mechanism mediated by nuclear factor erythroid 2 like 1 (NFE2L1) and aspartic protease DNA damage inducible 1 homolog 2 (DDI2). Following DDI2-mediated cleavage, NFE2L1 is able to induce transcription of virtually all proteasome subunit genes. Under normal condition, cleaved NFE2L1 is constantly degraded by proteasome, whereas in the presence of PIs, it accumulates and induces proteasome synthesis which in turn promotes the development of drug resistance towards PIs. Here, we report that Nelfinavir (NFV), an HIV protease inhibitor, can inhibit DDI2 activity directly. Inhibition of DDI2 by NFV effectively blocks NFE2L1 proteolysis and potentiates cytotoxicity of PIs in cancer cells. Recent clinical evidence indicated that NFV can effectively delay the refractory period of MM patients treated with PI-based therapy. Our finding hence provides a specific molecular mechanism for combinatorial therapy using NFV and PIs for treating MM and probably additional cancers.

Rotor speed estimation for induction motors is a key problem in speed-sensorless motor drives. This paper performs nonlinear high gain observer design based on the full-order model of the induction motor. Such an effort appears non-trivial due to the fact that the full-model at best admits locally a non-triangular observable form &#x0028 NTOF &#x0029, and its analytical representation in the NTOF can not be obtained. This paper proposes an approximate high gain estimation algorithm, which enjoys a constructive design, ease of tuning, and improved speed estimation and tracking performance. Experiments demonstrate the effectiveness of the proposed algorithm.

This work investigates energy-optimal motion planning for a class of multiaxis motion systems where the system dynamics are linear time-invariant and decoupled in each axis. Solving the problem in a reliable and efficient manner remains challenging owing to the presence of various constraints on control and states, nonconvexity in its cost function, and obstacles. This paper shows how the cost function can be convexified by considering the system dynamics, while decomposing decision variables to obtain a convex representation of collision avoidance constraints. With the convexified cost function and constraints, the original problem is decomposed into two quadratic programming (QP) problems. An alternating quadratic programming (AQP) algorithm is proposed to solve both the QP problems alternatingly and iteratively till convergence. Requiring an initial feasible trajectory as a guess, AQP necessarily converges to an energy-efficient solution that is homotopic to the initial guess. Under certain circumstances, AQP is guaranteed to produce a local optimum. Simulation demonstrates that AQP is computationally efficient and reliable while claiming comparable energy saving as the mixed-integer QP approach. Note to Practitioners —This paper presents an energy-optimal motion planning algorithm that can be easily implemented on a class of multiaxis motion systems. Main advantages of the proposed algorithm are: 1) it produces a trajectory resulting in lower but comparable energy efficiency as the global optimum; 2) it is guaranteed to provide an energy-efficient and constraint-compliant trajectory, and thus is reliable; 3) it requires a low computational load and can be deployed on a wide range of applications; and 4) its implementation is straightforward to any engineer with basic knowledge of numerical methods.

The rapidly growing use of lithium-ion batteries across various industries highlights the pressing issue of optimal charging control, as charging plays a crucial role in the health, safety, and life of batteries. The literature increasingly adopts model predictive control (MPC) to address this issue, taking advantage of its capability of performing optimization under constraints. However, the computationally complex online constrained optimization intrinsic to MPC often hinders real-time implementation. This article is thus proposed to develop a framework for real-time charging control based on explicit MPC (eMPC), exploiting its advantage in characterizing an explicit solution to an MPC problem, to enable real-time charging control. This article begins with the formulation of MPC charging based on a nonlinear equivalent circuit model. Then, multisegment linearization is conducted to the original model, and applying the eMPC design to the obtained linear models leads to a charging control algorithm. The proposed algorithm shifts the constrained optimization to offline by precomputing explicit solutions to the charging problem and expressing the charging law as piecewise affine functions. This drastically reduces not only the online computational costs in the control run but also the difficulty of coding. Extensive numerical simulation and experimental results verify the effectiveness of the proposed eMPC charging control framework and algorithm. The research results can potentially meet the needs for real-time battery management running on embedded hardware.

This paper presents a general space vector modulation (SVM) method for the modular multilevel converter (MMC). Compared with earlier modulation methods, the proposed SVM method not only utilizes the maximum level number (i.e., $2n+ 1$ , where $n$ is the number of submodules in the upper or lower arm of each phase) of output phase voltages, but also leads to an optimized control performance in terms of capacitor voltage balancing, circulating current suppression, and common-mode voltage reduction. The maximum level number is achieved by introducing a new equivalent circuit of the MMC, and the optimized control is obtained by selecting the optimal redundant switching states. Since the computational burden of the SVM scheme is independent of the voltage level number, the proposed method is well suited to the MMC with any number of submodules. Simulation and experimental results are presented to validate the proposed method.

This paper presents novel machine-learning-based methods for estimating the state of charge (SoC) of lithium-ion batteries, which use the Gaussian process regression (GPR) framework. The measured battery parameters, such as voltage, current, and temperature, are used as inputs for regular GPR, whereas the SoC estimate at the previous sample is fed back and incorporated into the input vector for recurrent GPR. The proposed methods consist of two parts. In the first part, training is performed wherein the optimal hyperparameters of a chosen kernel function are determined to model data properties. In the second part, online SoC estimation is carried out according to the trained model. One of the practical advantages of a GPR framework is to quantify estimation uncertainty and, hence, to enable reliability assessment of the battery SoC estimate. The performance of the proposed methods is evaluated by using a simulated dataset and two experimental datasets, one with constant and the other with dynamic charge and discharge currents. The simulations and experimental results show the superiority of the proposed methods in comparison to state-of-the-art techniques including a support vector machine, a relevance vector machine, and a neural network.

This paper presents a novel modular, reconfigurable battery energy storage system. The proposed design is characterized by a tight integration of reconfigurable power switches and DC/DC converters. This characteristic enables isolation of faulty cells from the system and allows fine power control for individual cells toward optimal system-level performance. An optimal power management approach is developed to extensively exploit the merits of the proposed design. Based on receding-horizon convex optimization, this approach aims to minimize the total power losses in charging/discharging while allocating the power in line with each cell's condition to achieve state-of-charge (SoC) and temperature balancing. By appropriate design, the approach manages to regulate the power of a cell across its full SoC range and guarantees the feasibility of the optimization problem. We perform extensive simulations and further develop a lab-scale prototype to validate the proposed system design and power management approach., This work is published in the IEEE Transactions on Transportation Electrification

In this paper, we propose a shared steering control strategy for semi-autonomous vehicles by taking into account the interaction between the driver and his/her vehicle. The copilot controller cooperates with the driver and help maintain the vehicle in the central position of a lane. Taking advantage of the small-gain theory, the design procedure does not rely on the perfect knowledge of the model and states for the driver. An adaptive dynamic programming method is introduced to develop a shared controller in real-time using online measurable input-output data from vehicle sensors. The efficacy of the proposed shared steering controller is demonstrated by computer-based simulations.

With high energy/power density, flexible and lightweight design, low self-discharge rates and long cycle life, lithium-ion (Li+) batteries have experienced a surging growth since being commercialized in the early 1990s [1]. They are dominant today in the consumer electronics sector. Due to continually declining manufacturing costs, they are also rapidly penetrating sectors such as the power grid, renewable energy, automotive, and aerospace, where largescale energy storage is needed. Looking into the future, the role of Li+ batteries will be further strengthened as a key energy-storage technology to support the progression of the world into the green energy era. However, their vulnerability to overcharge, overdischarge, and overheating can easily expose them to performance degradation, shortened cycle life, and even fire and explosion, thus raising many concerns about their deployment. These challenges have been driving a massive solution-seeking effort in various relevant research fields. Associated with this trend is the control-theory-enabled advancement of battery-management system (BMS) technologies.

A formalism based on Pontryagin's maximum principle is applied to determine the time-optimal protocol that drives a general initial state to a target state by a Hamiltonian with limited control, i.e., there is a single control field with bounded amplitude. The coupling between the bath and the qubit is modeled by a Lindblad master equation. Dissipation typically drives the system to the maximally mixed state; consequently, there generally exists an optimal evolution time beyond which the decoherence prevents the system from getting closer to the target state. For some specific dissipation channel, however, the optimal control can keep the system from the maximum entropy state for infinitely long. The conditions under which this specific situation arises are discussed in detail. The numerical procedure to construct the time-optimal protocol is described. In particular, the formalism adopted here can efficiently evaluate the time-dependent singular control which turns out to be crucial in controlling either an isolated or a dissipative qubit., 7 figures

A stochastic procedure is developed which allows one to express Pontryagin's maximum principle for dissipative quantum system solely in terms of stochastic wave functions. Time-optimal controls can be efficiently computed without computing the density matrix. Specifically, the proper dynamical update rules are presented for the stochastic costate variables introduced by Pontryagin's maximum principle and restrictions on the form of the terminal cost function are discussed. The proposed procedure is confirmed by comparing the results to those obtained from optimal control on Lindbladian dynamics. Numerically, the proposed formalism becomes time and memory efficient for large systems, and it can be generalized to describe non-Markovian dynamics., Comment: 4 figures

Parameter estimation is of foundational importance for various model-based battery management tasks, including charging control, state-of-charge estimation and aging assessment. However, it remains a challenging issue as the existing methods generally depend on cumbersome and time-consuming procedures to extract battery parameters from data. Departing from the literature, this paper sets the unique aim of identifying all the parameters offline in a one-shot procedure, including the resistance and capacitance parameters and the parameters in the parameterized function mapping from the state-of-charge to the open-circuit voltage. Considering the well-known Thevenin’s battery model, the study begins with the parameter identifiability analysis, showing that all the parameters are locally identifiable. Then, it formulates the parameter identification problem in a prediction-error-minimization framework. As the non-convexity intrinsic to the problem may lead to physically meaningless estimates, two methods are developed to overcome this issue. The first one is to constrain the parameter search within a reasonable space by setting parameter bounds, and the other adopts regularization of the cost function using prior parameter guess. The proposed identifiability analysis and identification methods are extensively validated through simulations and experiments.

Data-driven methods for learning near-optimal control policies through approximate dynamic programming (ADP) have garnered widespread attention. In this paper, we investigate how data-driven control methods can be leveraged to imbue near-optimal performance in a core component in modern factory systems: the electric motor drive. We apply policy iteration-based ADP to an induction motor model in order to construct a state feedback control policy for a given cost functional. Approximate error convergence properties of policy iteration methods imply that the learned control policy is near-optimal. We demonstrate that carefully selecting a cost functional and initial control policy yields a near-optimal control policy that outperforms both a baseline nonlinear control policy based on backstepping, as well as the initial control policy.

Grover's algorithm is one of the most famous algorithms which explicitly demonstrates how the quantum nature can be utilized to accelerate the searching process. In this work, Grover's quantum search problem is mapped to a time-optimal control problem. Resorting to Pontryagin's Minimum Principle we find that the time-optimal solution has the bang-singular-bang structure. This structure can be derived naturally, without integrating the differential equations, using the geometric control technique where Hamiltonians in the Schr\"odinger's equation are represented as vector fields. In view of optimal control, Grover's algorithm uses the bang-bang protocol to approximate the optimal protocol with a minimized number of bang-to-bang switchings to reduce the query complexity. Our work provides a concrete example how Pontryagin's Minimum Principle is connected to quantum computation, and offers insight into how a quantum algorithm can be designed., Comment: 4 figures

Many data-driven control methodologies depend on an initial stabilizing control policy, and subsequently use operational data to refine the initial policy in order to optimize closed-loop performance. For general dynamical systems, computing such an initial policy is non-trivial, and systematic methods for this task are not available in the current literature. In this paper, we propose a systematic framework for constructing stabilizing and/or constraint-enforcing control policies for a class of nonlinear systems based on archival data. Specifically, we study partially unmodeled systems whose nonlinearities satisfy local Lipschitz conditions. We employ kernel density estimation (KDE) to learn a local Lipschitz constant from archival data, and compute control policies by solving semidefinite programs that leverage matrix multipliers informed by the Lipschitz learner. We demonstrate the potential of our proposed methodology on a nonlinear system with unmodeled dynamics.

Battery parameter identification is emerging as an important topic due to the increasing use of battery energy storage. This paper studies parameter identification for the nonlinear double-capacitor (NDC) model for lithium-ion batteries, which is a new equivalent circuit model developed in the authors' previous work [1]. It is noticed that the NDC model has a structure similar to the Wiener system. From this Wiener perspective, this work builds a parameter identification approach for this model upon the well-known maximum a posteriori (MAP) estimation. The purpose of using MAP is to overcome the nonconvexity and local minima that can cause unphysical parameter estimates. A quasi-Newton-based method is utilized to accomplish the involved optimization procedure numerically. The proposed approach is the first one that we aware of exploits MAP for Wiener system identification. It also demonstrates significant effectiveness for accurate identification of the NDC model as validated through experiments.

This article proposes a new equivalent circuit model for rechargeable batteries by modifying a double-capacitor model in the literature. It is known that the original model can address the rate capacity effect and energy recovery effect inherent to batteries better than other models. However, it is a purely linear model and includes no representation of a battery’s nonlinear phenomena. Hence, this article transforms the original model by introducing a nonlinear-mapping-based voltage source and a serial RC circuit. The modification is justified by an analogy with the single-particle model. Two off-line parameter estimation approaches, termed 1.0 and 2.0, are designed for the new model to deal with the scenarios of constant-current and variable-current charging/discharging, respectively. In particular, the 2.0 approach proposes the notion of Wiener system identification based on the maximum a posteriori estimation, which allows all the parameters to be estimated in one shot while overcoming the nonconvexity or local minima issue to obtain physically reasonable estimates. Extensive experimental evaluation shows that the proposed model offers excellent accuracy and predictive capability. A comparison against the Rint and Thevenin models further points to its superiority. With high fidelity and low mathematical complexity, this model is beneficial for various real-time battery management applications.

The rapidly growing use of lithium-ion batteries across various industries highlights the pressing issue of optimal charging control. The literature increasingly adopts model predictive control (MPC) to perform charging control, taking advantage of its capability of performing optimization under constraints. However, the computationally complex online constrained optimization involved in MPC often hinders real-time implementation. This paper is thus motivated to develop a new charging control algorithm based on explicit MPC (eMPC). Leveraging the merits of eMPC, the new algorithm can shift the constrained optimization to offline by precomputing explicit solutions to an optimal charging control problem and expressing the control law as piecewise functions. This drastically reduces not only the online computational costs in the control run but also the difficulty to code the algorithm. Numerical simulation results verify the utility of the proposed charging control algorithm, which can potentially meet the needs for real-time battery management running on embedded hardware.

This paper proposes a simplified space vector modulation (SVM) scheme for multilevel converters. Compared with earlier SVM methods, the proposed scheme simplifies the detection of the nearest three vectors and the generation of switching sequences, and therefore is computationally more efficient. Particularly, for the first time, the proposed scheme achieves the same easy implementation as phase-voltage modulation techniques. Another superior characteristic of the proposed scheme over earlier methods is its potential for multiphase multilevel applications. The proposed scheme also offers the following significant advantages: 1) independence of the level number of the converter; 2) more degrees of freedom, i.e., redundant switching sequences and adjustable duty cycles, to optimize the switching patterns; and 3) no need for lookup tables or coordinate transformations. These advantages make the proposed scheme well suited to large level-number applications, such as modular multilevel converters and high voltage direct current systems. Simulation and experimental results verify this new concept.

Along with the forth industry revolution, implementing industrial control systems over mainstream wireless networks such as WirelessHART, WiFi, and cellular networks becomes necessary. Well-known challenges, such as uncertain time delays and packet drops, induced by networks have been intensively investigated from various perspectives: control synthesis, network design, or control and network co-design. The status quo is that industry remains hesitant to close the loop at the control-to-actuation side due to safety concerns. This work offers an alternative perspective to address the safety concern, by exploiting the design freedom of system architecture. Specifically, we present a smart actuation architecture, which deploys (1) a remote controller, which communicates with physical plant via wireless network, accounting for optimality, adaptation, and constraints by conducting computationally expensive operations; (2) a smart actuator, which co-locates with the physical plant, executing a local control policy and accounting for system safety in the view of network imperfections. Both the remote and the local controllers run at the same time scale and cooperate through an unreliable network. We propose a policy iteration-based procedure to co-design the local and remote controllers when the latter employs the model predictive control policy. Semi-global asymptotic stability of the resulting closed-loop system can be established for certain classes of plants. Extensive simulations demonstrate the advantages of the proposed architecture and co-design procedure.

Approximate/adaptive dynamic programming (ADP) has demonstrated great successes in the construction of data-driven output feedback optimal control for linear time-invariant systems and data-driven state feedback optimal control for nonlinear systems. This work investigates data-driven output feedback optimal control design for a class of nonlinear systems. It proposes to parameterize all admissible output feedback optimal control policies over accessible signals (system output and its time derivatives). In the case that system state can be parameterized as functions of accessible signals, then the value function and control policy can be parameterized over accessible signals, which allow ADP to be driven by accessible data. For a special case, where system state, value function and control policy can be linearly parameterized over a finite functional space over accessible signals, the policy iteration algorithm (PI) of ADP is reduced to solve a system of linear equations. Two data-driven PIs are developed to accomplish data-driven output feedback optimal control design. Simulation validates the proposed methodology.