In goal-conditioned hierarchical reinforcement learning (HRL), a high-level policy specifies a subgoal for the low-level policy to reach. Effective HRL hinges on a suitable subgoal representation function, abstracting...
详细信息
In goal-conditioned hierarchical reinforcement learning (HRL), a high-level policy specifies a subgoal for the low-level policy to reach. Effective HRL hinges on a suitable subgoal representation function, abstracting state space into latent subgoal space and inducing varied low-level behaviors. Existing methods adopt a subgoal representation that provides a deterministic mapping from state space to latent subgoal space. Instead, this paper utilizes Gaussian Processes (GPs) for the first probabilistic subgoal representation. Our method employs a GP prior on the latent subgoal space to learn a posterior distribution over the subgoal representation functions while exploiting the long-range correlation in the state space through learnable kernels. This enables an adaptive memory that integrates long-range subgoal information from prior planning steps allowing to cope with stochastic uncertainties. Furthermore, we propose a novel learning objective to facilitate the simultaneous learning of probabilistic subgoal representations and policies within a unified framework. In experiments, our approach outperforms state-of-the-art baselines in standard benchmarks but also in environments with stochastic elements and under diverse reward conditions. Additionally, our model shows promising capabilities in transferring low-level policies across different tasks. Copyright 2024 by the author(s)
The growing adoption of renewable energy primary plants has been seen globally. South Africa has seen a significant development in utility-scale photovoltaic (PV) farms in recent years due to the alleviated regulation...
详细信息
With the increase in the amount of tasks offtoaded to the network edge, the energy supply of edge devices has become a challenge worthy of attention. It is a feasible way to use renewable energy to supply energy for e...
详细信息
With the increase in the amount of tasks offtoaded to the network edge, the energy supply of edge devices has become a challenge worthy of attention. It is a feasible way to use renewable energy to supply energy for edge devices, but production of renewable energy has certain uncertainty and stochasticity. In order to provide sufficient energy to ensure stable operation of edge devices, energy Internet (EI) provides an idea, that is, different edge devices are connected with distributed small energy supply and storage systems. As the core equipment of energy Internet, energy router (ER) plays an important role in information transmission, energy transmission and system control. In this paper, the concept of edge energy router is proposed, which has the ability of task computing and scheduling similar to edge computing server, as well as the ability of energy transmission and system control of energy router. Each edge energy router is connected with loads, photovoltaic panel (PV), micro turbine (MT) and battery energy storage (BES) to form a self-sufficient microgrid (MG) system. However, there exists a delay in energy transmission and task scheduling between different ERs. Moreover, the DC bus voltage stability of each edge energy router system is negatively affected by internal uncertainty, stochasticity and external interference. Therefore, the system is modeled by Markov jump ODEs with time delay, and robust control of DC bus voltage deviation is discussed in this paper. The linear matrix inequality (LMI) method is used to solve this Markov jump control problem. Finally, numerical simulations show the effectiveness of the proposed method.
Smart cities require better cellular communication, incorporating a high data rate that satisfies the Internet, cloud computing, and the Internet of Things (IoT) requirements. A high data rate demands higher bandwidth...
详细信息
Visible light communication (VLC) has emerged as a cutting-edge high-speed communication technology, poised to meet the surging capacity demands of 6G networks. Micro-light-emitting diodes (μLEDs) are considered as t...
详细信息
Traffic light control plays a crucial role in intelligent transportation systems. This paper introduces Temporal Difference-Aware Graph Convolutional Reinforcement Learning (TeDA-GCRL), a decentralized RL-based method...
详细信息
In this study, we present high-performance photonic-crystal surface-emitting lasers (PCSELs) operating at short wavelength infrared (SWIR) band of approximately 1580 nm. These PCSELs demonstrate impressive output powe...
详细信息
This paper presents the design, fabrication, and characterization of a novel 3D-printed helical antenna operating within the 9.4–10.8 GHz frequency band. The antenna, employing a lightweight paper substrate and a str...
详细信息
With the increasing demand for power system stability and resilience,effective real-time tracking plays a crucial role in smart grid ***,most studies have focused on measurement noise,while they seldom think about the...
详细信息
With the increasing demand for power system stability and resilience,effective real-time tracking plays a crucial role in smart grid ***,most studies have focused on measurement noise,while they seldom think about the problem of measurement data loss in smart power grid *** solve this problem,a resilient fault-tolerant extended Kalman filter(RFTEKF)is proposed to track voltage amplitude,voltage phase angle and frequency ***,a threephase unbalanced network’s positive sequence fast estimation model is ***,the loss phenomenon of measurements occurs randomly,and the randomness of data loss’s randomness is defined by discrete interval distribution[0,1].Subsequently,a resilient fault-tolerant extended Kalman filter based on the real-time estimation framework is designed using the timestamp technique to acquire partial data loss ***,extensive simulation results manifest the proposed RFTEKF can synchronize the smart grid more effectively than the traditional extended Kalman filter(EKF).
Ultrafast 3D imaging is indispensable for visualizing complex and dynamic biological *** scanning-based techniques necessitate an inherent trade-off between acquisition speed and space-bandwidth product(SBP).Emerging ...
详细信息
Ultrafast 3D imaging is indispensable for visualizing complex and dynamic biological *** scanning-based techniques necessitate an inherent trade-off between acquisition speed and space-bandwidth product(SBP).Emerging single-shot 3D wide-field techniques offer a promising alternative but are bottlenecked by the synchronous readout constraints of conventional CMOS systems,thus restricting data throughput to maintain high SBP at limited frame *** address this,we introduce EventLFM,a straightforward and cost-effective system that overcomes these challenges by integrating an event camera with Fourier light field microscopy(LFM),a state-of-theart single-shot 3D wide-field imaging *** event camera operates on a novel asynchronous readout architecture,thereby bypassing the frame rate limitations inherent to conventional CMOS *** further develop a simple and robust event-driven LFM reconstruction algorithm that can reliably reconstruct 3D dynamics from the unique spatiotemporal measurements captured by *** results demonstrate that EventLFM can robustly reconstruct fast-moving and rapidly blinking 3D fluorescent samples at kHz frame ***,we highlight EventLFM’s capability for imaging of blinking neuronal signals in scattering mouse brain tissues and 3D tracking of GFP-labeled neurons in freely moving *** believe that the combined ultrafast speed and large 3D SBP offered by EventLFM may open up new possibilities across many biomedical applications.
暂无评论