Discovering Emergent Agent Behaviour with Evolutionary Finite State Machines

Mašek, Martin; Lam, Chiou Peng; Benke, Lyndon; Kelly, Luke J.; Papasimeon, Michael

doi:10.1007/978-3-030-03098-8_2

Cited by 12 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, ACE0 was specifically designed to accommodate different types of agent reasoning models. It has been used with a range of agent reasoning technologies, including automated planning (Ramirez et al, 2017(Ramirez et al, , 2018, evolutionary algorithms (Masek et al, 2018;Lam et al, 2019;Masek et al, 2021), reinforcement learning (Kurniawan et al, 2019(Kurniawan et al, , 2020 and Generative Adversarial Networks (Hossam et al, 2020). In this work, we build upon the work of (Ramirez et al, 2017(Ramirez et al, , 2018 using an automated hybrid planning approach combined with Model Predictive Control (MPC) to define and generate the behaviours in ACE0.…”

Section: Simulation Environmentmentioning

confidence: 99%

X-ray diffraction of photovoltaic perovskites: Principles and applications

Tan

McNeill

2022

Applied Physics Reviews

View full text Add to dashboard Cite

Solar cells based on organic–inorganic hybrid perovskite materials have emerged as the most efficient next-generation thin-film solar cells within just a decade of research and show great promise for commercialization. As control of the thin-film microstructure of the perovskite layer is a key factor enabling high photovoltaic efficiency, good stability, and successful up-scaling of high-quality perovskite thin films for commercialization, a reliable and accurate characterization of the thin-film microstructure is paramount. X-ray diffraction (XRD)-based techniques, including conventional laboratory-based XRD and synchrotron-based grazing-incidence wide-angle x-ray scattering, are widely used to probe the microstructure of photovoltaic perovskite thin films. Nevertheless, it is common for these XRD experiments to be poorly executed and diffraction data to be improperly interpreted. This review focuses on principles of XRD techniques and their application for the characterization of the perovskite thin-film microstructure. Fundamentals of XRD techniques are presented with a strong emphasis on best practices in data collection and analysis. Approaches for the reliable and accurate extraction of microstructural information from diffraction data are discussed, including the need for simulating diffraction patterns. Applications of XRD techniques in characterizing perovskite thin films are demonstrated for both three-dimensional and layered hybrid perovskites, covering various microstructural aspects including phase identification and quantification, texture analysis, microstrain, and macrostrain as well as in situ and operando characterization. The additional subtleties and complexities associated with the XRD characterization of layered hybrid perovskites due to a more complex thin-film microstructure are discussed. Common mistakes and pitfalls that lead to misinterpretation of diffraction data are also highlighted.

show abstract

Section: Simulation Environmentmentioning

confidence: 99%

X-ray diffraction of photovoltaic perovskites: Principles and applications

Tan

McNeill

2022

Applied Physics Reviews

View full text Add to dashboard Cite

show abstract

“…The second environment we test our methods with is an air combat simulator called Ace Zero, which was developed by Defence Science and Technology Group, Australia and used in [30,22,17,14,15]. We set the simulator for one-on-one fights in two-dimensional space, representing a continuous space sequential-decision problem much larger than Cartpole.…”

Section: The Environmentmentioning

confidence: 99%

Discrete-to-deep reinforcement learning methods

Kurniawan

Vamplew

Papasimeon

et al. 2021

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

Neural networks are effective function approximators, but hard to train in the reinforcement learning (RL) context mainly because samples are correlated. In complex problems a neural RL approach is often able to learn a better solution than tabular RL, but generally takes longer. This paper proposes two methods, Discrete-to-Deep Supervised Policy Learning (D2D-SPL) and Discrete-to-Deep Supervised Q-value Learning (D2D-SQL), whose objective is to acquire the generalisability of a neural network at a cost nearer to that of a tabular method. Both methods combine RL and supervised learning (SL) and are based on the idea that a fast-learning tabular method can generate off-policy data to accelerate learning in neural RL. D2D-SPL uses the data to train a classifier which is then used as a controller for the RL problem. D2D-SQL uses the data to initialise a neural network which is then allowed

show abstract

Section: Simulation Environmentmentioning

confidence: 99%

Behaviour Recognition with Kinodynamic Planning Over Continuous Domains

Fitzpatrick

Lipovetzky

Papasimeon

et al. 2021

Front. Artif. Intell.

Self Cite

View full text Add to dashboard Cite

We investigate the application of state-of-the-art goal recognition techniques for behaviour recognition over complex continuous domains using model predictive control (MPC) for trajectory generation. We formally define the problem of kinodynamic behaviour recognition and establish a set of baseline behaviours and performance measures in the complex domain of unmanned aerial maneuvers. We evaluate how well our approach performs over a range of standard aerial maneuvers and representative initial configurations of varying complexity. The work also highlights future research directions in compound model-based behaviour recognition and team behaviour recognition where multiple agents may be acting simultaneously.

show abstract

Discovering Emergent Agent Behaviour with Evolutionary Finite State Machines

Cited by 12 publications

References 8 publications

X-ray diffraction of photovoltaic perovskites: Principles and applications

X-ray diffraction of photovoltaic perovskites: Principles and applications

Discrete-to-deep reinforcement learning methods

Behaviour Recognition with Kinodynamic Planning Over Continuous Domains

Contact Info

Product

Resources

About