Results 141 to 150 of about 7,404,725 (288)
Counterfactually Safe Reinforcement Learning
Reinforcement learning algorithms are generally designed to maximize the expected return across a population. However, a policy that is optimal on average may be suboptimal for certain individuals, leading to potential safety concerns. To address this, we first formalize the notion of individual harm from a counterfactual perspective and define harm as
Li, Jingyi, Wu, Peng, Shi, Chengchun
openaire +2 more sources
Safe and Robust Reinforcement Learning: Principles and Practice
CoRRReinforcement Learning (RL) has shown remarkable success in solving relatively complex tasks, yet the deployment of RL systems in real-world scenarios poses significant challenges related to safety and robustness. This paper aims to identify and further understand those challenges thorough the exploration of the main dimensions of the safe and robust ...
Taku Yamagata, Raúl Santos-Rodríguez
openaire +3 more sources
Correction: Success-efficient/failure-safe strategy for hierarchical reinforcement motor learning.
PLoS Computational Biology[This corrects the article DOI: 10.1371/journal.pcbi.1013089.].
PLOS Computational Biology Staff
doaj +1 more source
Autonomous Navigation of Pollen‐Inspired Magnetic Microrobots for Biomedical Applications
Advanced Robotics Research, EarlyView.A sunflower pollen‐inspired magnetic microrobot enables controlled rolling navigation in vessel‐like environments. Its open geometry reduces hydrodynamic drag, while vision‐based closed‐loop control, shortest‐path planning, and reinforcement learning support target‐reaching and maze navigation with microrobot‐scale accuracy, highlighting a route toward
Ali Anil Demircali +6 more
wiley +1 more source
Flexible Sensors for Robotics Tactile Perception: A Review
Advanced Robotics Research, EarlyView.Flexible tactile sensing for robotics is reviewed through four interconnected dimensions. Physical mechanisms include piezoresistive, capacitive, piezoelectric, triboelectric, iontronic, and optical sensing. Structural design includes bioinspired, defect‐based, and MEMS‐based tactile systems.
Yu Song, Ying Chen, Yihao Chen, Xue Feng
wiley +1 more source
Kansas Safe Kids Annual Report
, 2022 Safe Kids Kansas is a 501c(3) charitable organization and is a member of Safe Kids Worldwide, a global organization dedicated to preventing injuries in children.
Safe Kids Kansas, Inc.
core
Explainably Safe Reinforcement Learning
Trust in a decision-making system requires both safety guarantees and the ability to interpret and understand its behavior. This is particularly important for learned systems, whose decision-making processes are often highly opaque. Shielding is a prominent model-based technique for enforcing safety in reinforcement learning.
Rieder, Sabine +4 more
openaire +2 more sources
Safe and Robust Binary Classification and Fault Detection Using Reinforcement Learning
IEEE Open Journal of Control SystemsIn this paper, we propose a learning-based method utilizing the Soft Actor-Critic (SAC) algorithm to train a binary Support Vector Machine (SVM) classifier.
Josh Netter +3 more
doaj +1 more source
Nonreciprocal Swarmalators With Reconfigurable and Controllable Formations for Robot Collectives
Advanced Robotics Research, EarlyView.Nonreciprocal swarmalator interactions are enabled through control barrier functions to transform self‐organizing robot collectives into reconfigurable, constraint‐aware systems. Complex two‐ and three‐dimensional shapes, continuous morphing, obstacle‐aware navigation, collective splitting, and object transport emerge from modulating agent‐level ...
Kush Patel +3 more
wiley +1 more source
Multi-agent reinforcement learning for planning and scheduling multiple goals
, 2000 Recently, reinforcement learning has been proposed as an effective method for knowledge acquisition of the multiagent systems. However, most researches on multiagent system applying a reinforcement learning algorithm focus on the method to reduce ...
Arai, Sachiyo +2 more
core +1 more source