Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2025.106021
HongSheng QI
Vehicle safety is crucial for the development of autonomous vehicles. The concept of RSS (responsibility sensitive safety) is proposed as a formal method to guarantee safety, and several longitudinal RSS formulations are proposed. However, even for highly accurate autonomous vehicles, its dynamics is stochastic, and may lead to the RSS violation due to disturbances. This research investigates the stochasticity of the adaptive cruise control empirically, and proposes a robust RSS regarding longitudinal driving. We firstly show that there exists certain time-variant stochasticity in the ACC controlled vehicle using real world dataset, and numerically calculated the probability that the RSS may be violated when the noise is considered, which reveals that RSS violation is inevitable under certain scenarios. Then, an explicit solution of the noise disturbed ACC is derived. Next, a robust RSS framework is established, by employing the Markov chain derived from the Euler-Maruyama scheme of the stochastic ACC system. The research extends the conventional RSS framework.
{"title":"Robust responsibility-sensitive safety: Noise disturbed adaptive cruise control","authors":"HongSheng QI","doi":"10.1016/j.sysconle.2025.106021","DOIUrl":"10.1016/j.sysconle.2025.106021","url":null,"abstract":"<div><div>Vehicle safety is crucial for the development of autonomous vehicles. The concept of RSS (responsibility sensitive safety) is proposed as a formal method to guarantee safety, and several longitudinal RSS formulations are proposed. However, even for highly accurate autonomous vehicles, its dynamics is stochastic, and may lead to the RSS violation due to disturbances. This research investigates the stochasticity of the adaptive cruise control empirically, and proposes a robust RSS regarding longitudinal driving. We firstly show that there exists certain time-variant stochasticity in the ACC controlled vehicle using real world dataset, and numerically calculated the probability that the RSS may be violated when the noise is considered, which reveals that RSS violation is inevitable under certain scenarios. Then, an explicit solution of the noise disturbed ACC is derived. Next, a robust RSS framework is established, by employing the Markov chain derived from the Euler-Maruyama scheme of the stochastic ACC system. The research extends the conventional RSS framework.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 106021"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143183651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.105988
Xiaojuan Li
In this paper, we study a stochastic optimal control problem under degenerate -expectation. By using implied partition method, we show that the approximation result for admissible controls still hold. Based on this result, we obtain the dynamic programming principle, and prove that the value function is the unique viscosity solution to the related HJB equation under degenerate case.
{"title":"Dynamic programming principle for stochastic optimal control problem under degenerate G-expectation","authors":"Xiaojuan Li","doi":"10.1016/j.sysconle.2024.105988","DOIUrl":"10.1016/j.sysconle.2024.105988","url":null,"abstract":"<div><div>In this paper, we study a stochastic optimal control problem under degenerate <span><math><mi>G</mi></math></span>-expectation. By using implied partition method, we show that the approximation result for admissible controls still hold. Based on this result, we obtain the dynamic programming principle, and prove that the value function is the unique viscosity solution to the related HJB equation under degenerate case.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 105988"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143183240","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.105985
Bin Zhang , Yingying Ren , Cuijuan An , Da-Wei Ding
This paper investigates the optimal tracking performance (OTP) of networked control systems (NCSs) with time delay subjected to multiplicative false data injection (FDI) attacks. An auxiliary model is proposed to preserve the influences of the first and second moments of the attacked channel output through a supplementary additive process. By employing a two-degree-of-freedom control (TDOF) scheme and a frequency-domain analysis approach, a display expression for the OTP of NCSs is derived. It is shown that the OTP has a close relation with the intrinsic characteristic of the plant, time delay, and statistical characteristics of multiplicative false data. Finally, a numerical example is provided to validate the results.
{"title":"Optimal tracking performance of networked control systems with time delay under multiplicative false data injection attacks","authors":"Bin Zhang , Yingying Ren , Cuijuan An , Da-Wei Ding","doi":"10.1016/j.sysconle.2024.105985","DOIUrl":"10.1016/j.sysconle.2024.105985","url":null,"abstract":"<div><div>This paper investigates the optimal tracking performance (OTP) of networked control systems (NCSs) with time delay subjected to multiplicative false data injection (FDI) attacks. An auxiliary model is proposed to preserve the influences of the first and second moments of the attacked channel output through a supplementary additive process. By employing a two-degree-of-freedom control (TDOF) scheme and a frequency-domain analysis approach, a display expression for the OTP of NCSs is derived. It is shown that the OTP has a close relation with the intrinsic characteristic of the plant, time delay, and statistical characteristics of multiplicative false data. Finally, a numerical example is provided to validate the results.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 105985"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143183239","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.106011
Alexander von Rohr, Dmitrii Likhachev, Sebastian Trimpe
We propose a data-driven control method for systems with aleatoric uncertainty, such as robot fleets with variations between agents. Our method leverages shared trajectory data to increase the robustness of the designed controller and thus facilitate transfer to new variations without the need for prior parameter and uncertainty estimation. In contrast to existing work on experience transfer for performance, our approach focuses on robustness and uses data collected from multiple realizations to guarantee generalization to unseen ones. Our method is based on scenario optimization combined with recent formulations for direct data-driven control. We derive upper bounds on the minimal amount of data required to provably achieve quadratic stability for probabilistic systems with aleatoric uncertainty and demonstrate the benefits of our data-driven method through a numerical example. We find that the learned controllers generalize well to high variations in the dynamics even when based on only a few short open-loop trajectories. Robust experience transfer enables the design of safe and robust controllers that work “out of the box” without additional learning during deployment.
{"title":"Robust direct data-driven control for probabilistic systems","authors":"Alexander von Rohr, Dmitrii Likhachev, Sebastian Trimpe","doi":"10.1016/j.sysconle.2024.106011","DOIUrl":"10.1016/j.sysconle.2024.106011","url":null,"abstract":"<div><div>We propose a data-driven control method for systems with aleatoric uncertainty, such as robot fleets with variations between agents. Our method leverages shared trajectory data to increase the robustness of the designed controller and thus facilitate transfer to new variations without the need for prior parameter and uncertainty estimation. In contrast to existing work on experience transfer for performance, our approach focuses on robustness and uses data collected from multiple realizations to guarantee generalization to unseen ones. Our method is based on scenario optimization combined with recent formulations for direct data-driven control. We derive upper bounds on the minimal amount of data required to provably achieve quadratic stability for probabilistic systems with aleatoric uncertainty and demonstrate the benefits of our data-driven method through a numerical example. We find that the learned controllers generalize well to high variations in the dynamics even when based on only a few short open-loop trajectories. Robust experience transfer enables the design of safe and robust controllers that work “out of the box” without additional learning during deployment.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 106011"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143183650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.106001
Thiem V. Pham, Quynh T. Thanh Nguyen
This article addresses the attacker’s perspective on an induced attack strategy against wide-area networks of clustered multi-agent systems. In contrast to most existing attacks, our suggested approach may force the wide-area network to follow a specified coordination trajectory. A hybrid ecosystem is first introduced for the induced attack model, which allows us to generate a particular synchronization trajectory. Next, an induced attack protocol is proposed that utilizes the induced attack model among partial agents and a dynamic event-triggered technique. It shows that the induced attack protocol is fully distributed, with each agent using only local information and that of its neighbors. An explicit expression is derived for the specified synchronization trajectory, which characterizes the movement trajectory of the wide-area network of clustered multi-agent systems under the induced attack. The resilient approach is used to provide design criteria for the induced attack generation ecosystem. Finally, simulation examples are provided to illustrate the practical implementation of the proposed methodology.
{"title":"Induced attack on synchronization trajectory against a wide-area network with hybrid event-triggered communication","authors":"Thiem V. Pham, Quynh T. Thanh Nguyen","doi":"10.1016/j.sysconle.2024.106001","DOIUrl":"10.1016/j.sysconle.2024.106001","url":null,"abstract":"<div><div>This article addresses the attacker’s perspective on an induced attack strategy against wide-area networks of clustered multi-agent systems. In contrast to most existing attacks, our suggested approach may force the wide-area network to follow a specified coordination trajectory. A hybrid ecosystem is first introduced for the induced attack model, which allows us to generate a particular synchronization trajectory. Next, an induced attack protocol is proposed that utilizes the induced attack model among partial agents and a dynamic event-triggered technique. It shows that the induced attack protocol is fully distributed, with each agent using only local information and that of its neighbors. An explicit expression is derived for the specified synchronization trajectory, which characterizes the movement trajectory of the wide-area network of clustered multi-agent systems under the induced attack. The resilient <span><math><msub><mrow><mi>H</mi></mrow><mrow><mi>∞</mi></mrow></msub></math></span> approach is used to provide design criteria for the induced attack generation ecosystem. Finally, simulation examples are provided to illustrate the practical implementation of the proposed methodology.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 106001"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143183643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper considers target tracking based on a beacon signal’s time-difference-of-arrival (TDOA) to a group of cooperating sensors. The sensors receive a reflected signal from the target where the time-of-arrival (TOA) renders the distance information. The existing approaches include: (i) classic centralized solutions which gather and process the target data at a central unit, (ii) distributed solutions which assume that the target data is observable in the dense neighborhood of each sensor (to be filtered locally), and (iii) double time-scale distributed methods with high rates of communication/consensus over the network. This work, in order to reduce the network connectivity in (i)-(ii) and communication rate in (iii), proposes a distributed single time-scale technique, which can also handle heterogeneous constant data-exchange delays over the static sensor network. This work assumes only distributed observability (in contrast to local observability in some existing works categorized in (ii)), i.e., the target is observable globally over a (strongly) connected network. The (strong) connectivity further allows for survivable network and -redundant observer design. Each sensor locally shares information and processes the received data in its immediate neighborhood via local linear-matrix-inequalities (LMI) feedback gains to ensure tracking error stability. The same gain matrix works in the presence of heterogeneous delays with no need of redesigning algorithms. Since most existing distributed estimation scenarios are linear (based on consensus), many works use linearization of the existing nonlinear TDOA measurement models where the output matrix is a function of the target position. As the exact target position is unknown, the existing works use estimated position in the output matrix (and for the gain design) at every time step. This makes their algorithm more complex and less accurate. Instead, this work provides a modified linear TDOA measurement model with a constant output matrix that is independent of target position and more practical in distributed linear setups.
{"title":"Distributed target tracking based on localization with linear time-difference-of-arrival measurements: A delay-tolerant networked estimation approach","authors":"Mohammadreza Doostmohammadian , Themistoklis Charalambous","doi":"10.1016/j.sysconle.2024.106009","DOIUrl":"10.1016/j.sysconle.2024.106009","url":null,"abstract":"<div><div>This paper considers target tracking based on a beacon signal’s time-difference-of-arrival (TDOA) to a group of cooperating sensors. The sensors receive a reflected signal from the target where the time-of-arrival (TOA) renders the distance information. The existing approaches include: (i) classic centralized solutions which gather and process the target data at a central unit, (ii) distributed solutions which assume that the target data is observable in the dense neighborhood of each sensor (to be filtered locally), and (iii) double time-scale distributed methods with high rates of communication/consensus over the network. This work, in order to reduce the network connectivity in (i)-(ii) and communication rate in (iii), proposes a distributed single time-scale technique, which can also handle heterogeneous constant data-exchange delays over the static sensor network. This work assumes only <em>distributed observability</em> (in contrast to local observability in some existing works categorized in (ii)), i.e., the target is observable globally over a (strongly) connected network. The (strong) connectivity further allows for <em>survivable network</em> and <span><math><mi>q</mi></math></span><em>-redundant observer design</em>. Each sensor locally shares information and processes the received data in its immediate neighborhood via local linear-matrix-inequalities (LMI) feedback gains to ensure tracking error stability. The same gain matrix works in the presence of heterogeneous delays with no need of redesigning algorithms. Since most existing distributed estimation scenarios are linear (based on consensus), many works use <em>linearization</em> of the existing <em>nonlinear TDOA measurement models</em> where the output matrix is a function of the target position. As the exact target position is unknown, the existing works use <em>estimated</em> position in the output matrix (and for the gain design) at every time step. This makes their algorithm more complex and less accurate. Instead, this work provides a <em>modified</em> linear TDOA measurement model with a <em>constant</em> output matrix that is independent of target position and more practical in distributed linear setups.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 106009"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182991","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.105986
Li Qiu, Lingtao Dong, Zhixing Shao, Shuyou Zhou
The coexistence of network denial of service (DoS) attacks and trigger transmission mechanism poses a challenge to the stability and control accuracy of semi-Markov jump systems (S-MJSs). DoS attacks with random characteristics and event-triggered mechanisms with dynamic parameters are studied in this paper. The communication topology jump process between multiple sub-actuators is modeled as a semi-Markov chain, and a S-MJSs with dual-channel DoS attacks and dynamic event-triggered mechanism (DETM) of sensor-to-controller (S-C) channel is obtained. In order to match the predictive compensation sequence with the trigger time of the S-C channel, a dynamic event-triggered predictive control (ETPC) strategy is designed by combining the controller with the trigger. In addition, the dynamic ETPC strategy proposed in this paper is applied to the tracking control of multi-motor, which effectively guarantees the tracking performance of multi-motor. Finally, the feasibility and superiority of the control strategy are verified by comparing with the network predictive control (NPC) strategy and the static ETPC strategy.
{"title":"Dynamic event-triggered predictive control strategy for networked semi-Markov jump systems","authors":"Li Qiu, Lingtao Dong, Zhixing Shao, Shuyou Zhou","doi":"10.1016/j.sysconle.2024.105986","DOIUrl":"10.1016/j.sysconle.2024.105986","url":null,"abstract":"<div><div>The coexistence of network denial of service (DoS) attacks and trigger transmission mechanism poses a challenge to the stability and control accuracy of semi-Markov jump systems (S-MJSs). DoS attacks with random characteristics and event-triggered mechanisms with dynamic parameters are studied in this paper. The communication topology jump process between multiple sub-actuators is modeled as a semi-Markov chain, and a S-MJSs with dual-channel DoS attacks and dynamic event-triggered mechanism (DETM) of sensor-to-controller (S-C) channel is obtained. In order to match the predictive compensation sequence with the trigger time of the S-C channel, a dynamic event-triggered predictive control (ETPC) strategy is designed by combining the controller with the trigger. In addition, the dynamic ETPC strategy proposed in this paper is applied to the tracking control of multi-motor, which effectively guarantees the tracking performance of multi-motor. Finally, the feasibility and superiority of the control strategy are verified by comparing with the network predictive control (NPC) strategy and the static ETPC strategy.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 105986"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.106005
Nan Li , Ilya Kolmanovsky , Hong Chen
In this paper, we propose a novel data-driven moving-horizon control approach for systems subject to time-domain constraints. The approach combines the strengths of control for rejecting disturbances and MPC for handling constraints. In particular, the approach can dynamically adapt disturbance attenuation performance depending on measured system state and forecasted disturbance level to satisfy constraints. We establish theoretical properties of the approach including robust guarantees of closed-loop stability, disturbance attenuation, constraint satisfaction under noisy data, as well as sufficient conditions for recursive feasibility, and illustrate the approach with a numerical example.
{"title":"Data-driven moving-horizon control with adaptive disturbance attenuation for constrained systems","authors":"Nan Li , Ilya Kolmanovsky , Hong Chen","doi":"10.1016/j.sysconle.2024.106005","DOIUrl":"10.1016/j.sysconle.2024.106005","url":null,"abstract":"<div><div>In this paper, we propose a novel data-driven moving-horizon control approach for systems subject to time-domain constraints. The approach combines the strengths of <span><math><msub><mrow><mi>H</mi></mrow><mrow><mi>∞</mi></mrow></msub></math></span> control for rejecting disturbances and MPC for handling constraints. In particular, the approach can dynamically adapt <span><math><msub><mrow><mi>H</mi></mrow><mrow><mi>∞</mi></mrow></msub></math></span> disturbance attenuation performance depending on measured system state and forecasted disturbance level to satisfy constraints. We establish theoretical properties of the approach including robust guarantees of closed-loop stability, disturbance attenuation, constraint satisfaction under noisy data, as well as sufficient conditions for recursive feasibility, and illustrate the approach with a numerical example.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 106005"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.106004
Yingkai Song, Kamil A. Khan
Deterministic methods for global optimization typically rely on convex relaxations to infer crucial bounding information. In global dynamic optimization, comparing convex relaxation approaches can be challenging due to limitations in established differential inequality theory. In this article, we provide new comparison results for the Carathéodory solutions of related ordinary differential equations (ODEs). Our results are applicable to certain ODEs whose right-hand side functions need not be quasimonotonic or continuous with respect to state variables; they require only a weakened variant of the standard Lipschitz continuity assumption, along with mild differential inequality requirements motivated by interval analysis. Using our new comparison results, we reveal an intuitive monotonicity result for global dynamic optimization that was previously unknown: when constructing convex relaxations of parametric ODE solutions within a relaxation framework by Scott and Barton (J. Glob. Optim. 57:143–176, 2013), supplying tighter relaxations of the ODE right-hand sides will always translate into relaxations of the ODE solutions that are at least as tight.
{"title":"New inequalities for comparing ordinary differential equations arising in global dynamic optimization","authors":"Yingkai Song, Kamil A. Khan","doi":"10.1016/j.sysconle.2024.106004","DOIUrl":"10.1016/j.sysconle.2024.106004","url":null,"abstract":"<div><div>Deterministic methods for global optimization typically rely on convex relaxations to infer crucial bounding information. In global dynamic optimization, comparing convex relaxation approaches can be challenging due to limitations in established differential inequality theory. In this article, we provide new comparison results for the Carathéodory solutions of related ordinary differential equations (ODEs). Our results are applicable to certain ODEs whose right-hand side functions need not be quasimonotonic or continuous with respect to state variables; they require only a weakened variant of the standard Lipschitz continuity assumption, along with mild differential inequality requirements motivated by interval analysis. Using our new comparison results, we reveal an intuitive monotonicity result for global dynamic optimization that was previously unknown: when constructing convex relaxations of parametric ODE solutions within a relaxation framework by Scott and Barton (J. Glob. Optim. 57:143–176, 2013), supplying tighter relaxations of the ODE right-hand sides will always translate into relaxations of the ODE solutions that are at least as tight.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 106004"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182993","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-01DOI: 10.1016/j.sysconle.2024.105972
Vivek S. Borkar
Using the framework of Markov decision processes (equivalently, controlled Markov chains) as a test case, we point out some non-robustness issues that arise in small noise limits in stochastic control. This is used to motivate a novel formulation of multi-chain Markov decision processes.
{"title":"Some big issues with small noise limits in Markov decision processes","authors":"Vivek S. Borkar","doi":"10.1016/j.sysconle.2024.105972","DOIUrl":"10.1016/j.sysconle.2024.105972","url":null,"abstract":"<div><div>Using the framework of Markov decision processes (equivalently, controlled Markov chains) as a test case, we point out some non-robustness issues that arise in small noise limits in stochastic control. This is used to motivate a novel formulation of multi-chain Markov decision processes.</div></div>","PeriodicalId":49450,"journal":{"name":"Systems & Control Letters","volume":"196 ","pages":"Article 105972"},"PeriodicalIF":2.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143182996","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号