Intelligent wind farm control via deep reinforcement learning and high-fidelity simulations

AbstractHigh-fidelity simulations of coughs and sneezes that serve as virtual experiments are presented, and they offer an unprecedented opportunity to peer into the chaotic evolution of the resulting airborne droplet clouds. While larger droplets quickly fall-out of the cloud, smaller droplets evaporate rapidly. The non-volatiles remain airborne as droplet nuclei for a long time to be transported over long distances. The substantial variation observed between the different realizations has important social distancing implications, since probabilistic outlier-events do occur and may need to be taken into account when assessing the risk of contagion. Contrary to common expectations, we observe dry ambient conditions to increase by more than four times the number of airborne potentially virus-laden nuclei, as a result of reduced droplet fall-out through rapid evaporation. The simulation results are used to validate and calibrate a comprehensive multiphase theory, which is then used to predict the spread of airborne nuclei under a wide variety of ambient conditions.

Download Full-text

The Objective Structured Clinical Examination (Osce) In High-Fidelity Simulations for Assessing Nursing Students' Clinical Judgment

Proceedings of the 2019 3rd International Conference on Education and Multimedia Technology - ICEMT 2019 ◽

10.1145/3345120.3345127 ◽

2019 ◽

Author(s):

Hao Bin Yuan

Keyword(s):

Nursing Students ◽

Clinical Examination ◽

Clinical Judgment ◽

Objective Structured Clinical Examination ◽

High Fidelity ◽

High Fidelity Simulations

Download Full-text

High-fidelity simulations of gravity currents using a high-order finite-difference spectral vanishing viscosity approach

Computers & Fluids ◽

10.1016/j.compfluid.2021.104902 ◽

2021 ◽

pp. 104902

Author(s):

Ricardo A.S. Frantz ◽

Georgios Deskos ◽

Sylvain Laizet ◽

Jorge H. Silvestrini

Keyword(s):

Finite Difference ◽

Gravity Currents ◽

High Order ◽

High Fidelity ◽

Vanishing Viscosity ◽

High Fidelity Simulations ◽

High Order Finite Difference

Download Full-text

Transfer Reinforcement Learning for Autonomous Driving

ACM Transactions on Modeling and Computer Simulation ◽

10.1145/3449356 ◽

2021 ◽

Vol 31 (3) ◽

pp. 1-26

Author(s):

Aravind Balakrishnan ◽

Jaeyoung Lee ◽

Ashish Gaurav ◽

Krzysztof Czarnecki ◽

Sean Sedwards

Keyword(s):

Decision Making ◽

Reinforcement Learning ◽

Transfer Problem ◽

Autonomous Driving ◽

High Fidelity ◽

Rule Based ◽

High Level ◽

Real Vehicle

Reinforcement learning (RL) is an attractive way to implement high-level decision-making policies for autonomous driving, but learning directly from a real vehicle or a high-fidelity simulator is variously infeasible. We therefore consider the problem of transfer reinforcement learning and study how a policy learned in a simple environment using WiseMove can be transferred to our high-fidelity simulator, W ise M ove . WiseMove is a framework to study safety and other aspects of RL for autonomous driving. W ise M ove accurately reproduces the dynamics and software stack of our real vehicle. We find that the accurately modelled perception errors in W ise M ove contribute the most to the transfer problem. These errors, when even naively modelled in WiseMove , provide an RL policy that performs better in W ise M ove than a hand-crafted rule-based policy. Applying domain randomization to the environment in WiseMove yields an even better policy. The final RL policy reduces the failures due to perception errors from 10% to 2.75%. We also observe that the RL policy has significantly less reliance on velocity compared to the rule-based policy, having learned that its measurement is unreliable.

Download Full-text