Results 281 to 290 of about 1,163,868 (343)

Multiple Unmanned Aerial Vehicle Formation Control through Deep Reinforcement Learning with Offline Sample Correction

open access: yesAdvanced Intelligent Systems, EarlyView.
Herein, a deep reinforcement learning‐based multi‐UAV formation control approach is proposed. By optimizing the utilization of historical data through correcting of offline samples, the past experience is better leveraged and learning performance is improved.
Zhongkai Chen   +4 more
wiley   +1 more source

The Extension curriculum – teaching and learning experience of the extension project Agile methodologies for the sustainability of micro and small companies at the SENAC University Center

open access: gold
Milena da Silva de Paiva   +5 more
openalex   +1 more source

Dynamic Distribution‐Based Multiagent Cooperative Defense: An Enhanced Resource Allocation and Defense Efficiency Strategy

open access: yesAdvanced Intelligent Systems, EarlyView.
This study addresses multiagent defense challenges against low‐cost swarm attacks through a hierarchical framework combining resource allocation and PPO optimization. The three‐layer architecture coordinates strategic planning, deployment optimization, and real‐time execution.
Xiaokai Fei   +2 more
wiley   +1 more source

Insect‐Inspired Resilient Machines

open access: yesAdvanced Intelligent Systems, EarlyView.
This study presents a decentralized resilient control for legged robots, enabling self‐organized locomotion and rapid adaptation to extreme leg loss within seconds. It encodes the self‐embodied resilience strategies observed in stick insects and relies on neural dynamics with synaptic plasticity, minimal sensory feedback, and dynamic robot–environment ...
Thirawat Chuthong   +3 more
wiley   +1 more source

Harnessing Nonidealities in Analog In‐Memory Computing Circuits: A Physical Modeling Approach for Neuromorphic Systems

open access: yesAdvanced Intelligent Systems, EarlyView.
This work harnesses nonidealities in analog in‐memory computing (IMC) by training physical neural networks modeled with ordinary differential equations. A differentiable spike‐time discretization accelerates training by 20× and reduces memory usage by 100×, enabling large IMC‐equivalent models to learn the CIFAR‐10 dataset.
Yusuke Sakemi   +5 more
wiley   +1 more source

Home - About - Disclaimer - Privacy