Ensuring safety is a crucial challenge when deploying reinforcement learning (RL) to real-world systems. We develop confidence-based safety filters, a control-theoretic approach for certifying state safety constraints...
详细信息
ISBN:
(数字)9781665467612
ISBN:
(纸本)9781665467629
Ensuring safety is a crucial challenge when deploying reinforcement learning (RL) to real-world systems. We develop confidence-based safety filters, a control-theoretic approach for certifying state safety constraints for nominal policies learnt via standard RL techniques, based on probabilistic dynamics models. Our approach is based on a reformulation of state constraints in terms of cost functions, reducing safety verification to a standard RL task. By exploiting the concept of hallucinating inputs, we extend this formulation to determine a "backup" policy which is safe for the unknown system with high probability. The nominal policy is minimally adjusted at every time step during a roll-out towards the backup policy, such that safe recovery can be guaranteed afterwards. We provide formal safety guarantees, and empirically demonstrate the effectiveness of our approach.
In order to further understand the mechanism of material volume change in the drying process,numerical simulations(considering or neglecting shrinkage)of heat and mass transfer during convective drying of carrot slice...
详细信息
In order to further understand the mechanism of material volume change in the drying process,numerical simulations(considering or neglecting shrinkage)of heat and mass transfer during convective drying of carrot slices under constant and controlled temperature and relative humidity were carried *** results were validated with experimental *** results of the simulation show that the Quadratic model fitted well to the moisture ratio and the material temperature data trend with average relative errors of 5.9%and 8.1%,***,the results of the simulation considering shrinkage show that the moisture and temperature distributions during drying are closer to the experimental data than the results of the simulation disregarding *** material moisture content was significantly related to the shrinkage of dried *** and relative humidity significantly affected the volume shrinkage of carrot *** volume shrinkage increased with the rising of the constant temperature and the decline of relative *** model can be used to provide more information on the dynamics of heat and mass transfer during drying and can also be adapted to other products and dryers devices.
A procedure for the postoptimal analysis of dynamic positioning control system of floating vessels is proposed. The control system design is based on the optimal constrained covariance control (OC 3 ). Using the OC 3 ...
详细信息
A procedure for the postoptimal analysis of dynamic positioning control system of floating vessels is proposed. The control system design is based on the optimal constrained covariance control (OC 3 ). Using the OC 3 technique, the disadvantages of the classical LQG optimal control technique are avoided. The presented numerical example illustrates the properties of the new approach.
A structure-based image similarity measurement called DTWT-SSIM is presented. The main idea behind DTWT-SSIM is to combine the shift-invariance advantage of dual-tree wavelet transform (DTWT) with the structure-preser...
详细信息
A structure-based image similarity measurement called DTWT-SSIM is presented. The main idea behind DTWT-SSIM is to combine the shift-invariance advantage of dual-tree wavelet transform (DTWT) with the structure-preserving property of the structural similarity metrics (SSIM). A series of experimental results show the improved measurement to be an effective and stable metric in the comparison of edge maps when small noise and distortion appear in the images.
Unmanned Ariel Vehicles (UAVs) have gained significant importance in diverse sectors. Thus, a profound safety risk analysis/assessment to prevent any possible damage to themselves, the environment, and humans is funda...
详细信息
Unmanned Ariel Vehicles (UAVs) have gained significant importance in diverse sectors. Thus, a profound safety risk analysis/assessment to prevent any possible damage to themselves, the environment, and humans is fundamental for building and utilizing UAVs. To achieve that, two fundamental challenges should be addressed: i) identification of types and frequency of the issues and ii) assessment of their impact. In this paper, we aim to address the first challenge by automatizing the process of data field analysis. To do so, we first performed some statistical analysis of the reported issues of UAV systems (in Github) and manually extracted detailed data from the reports to better understand the type and nature of the issues. Then, to automatize the analysis, we used natural language processing algorithm to extract the keywords from the reports, and then applied four machine learning algorithms to build classifier models to classify the reports according to the fault category and severity level. The good performance results obtained suggest that these analyzes can be performed to further understand the UAV system issues, and help in the risk assessment procedure to identify the hazard and define the frequency and severity of the risk. Moreover, the results of this work can help a big community of developers and researchers in the precise and fast analysis of bug reports and safety risk assessment of any software system.
This paper considers the problem of using an integral sliding mode strategy to reduce the disturbance terms acting on nonlinear systems in regular form. It is proved that the definition of a suitable sliding manifold ...
详细信息
An improved active disturbance rejection controller based on the fractional order extended state observer（FOESO）is proposed in this *** the proposed FOESO,a second order plant model is converted into a cascaded frac...
详细信息
An improved active disturbance rejection controller based on the fractional order extended state observer（FOESO）is proposed in this *** the proposed FOESO,a second order plant model is converted into a cascaded fractional order integrator(1/s,0<α<1).Thus,a stable closed-loop feedback control system with enough phase margin for stability can be realized using a simple proportional ***,the open-loop phase-frequency characteristic of the system is flat around the gain crossover frequency,namely,the system is robust to loop gain ***,without the differential action in the designed controller,the control system achieves the robustness to high-frequency noise.
Random matrix theory has proven to be a very instrumental tool for the computation of the average capacity of MIMO communication channels. The problem formulation has been limited to Rayleigh fading, which makes the r...
详细信息
In this paper, the power quality of interconnected microgrids is managed using a Model Predictive control (MPC) methodology which manipulates the power converters of the microgrids in order to achieve the requirements...
详细信息
In this paper, the power quality of interconnected microgrids is managed using a Model Predictive control (MPC) methodology which manipulates the power converters of the microgrids in order to achieve the requirements. The control algorithm is developed for the microgrids working modes: grid-connected, islanded and interconnected. The results and simulations are also applied to the transition between the different working modes. In order to show the potential of the control algorithm a comparison study is carried out with classical Proportional-Integral Pulse Width Modulation (PI-PWM) based controllers. The proposed control algorithm not only improves the transient response in comparison with classical methods but also shows an optimal behavior in all the working modes, minimizing the harmonics content in current and voltage even with the presence of non-balanced and non-harmonic-free three-phase voltage and current systems.
In this study, we develop an immersed boundary method - volume of fluid (IBM-VOF) two-phase flow solver to simulate two-phase flow problem contains solid boundaries and free surface and use it to solve the typical pro...
详细信息
暂无评论