Sampling-Based Robust Control of Autonomous Systems with Non-Gaussian Noise

Authors: Thom S. Badings, Alessandro Abate, Nils Jansen, David Parker, Hasan A. Poonawala, Marielle Stoelinga9669-9678

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Realistic benchmarks show the practical applicability of our method, even when the i MDP has millions of states or transitions. We implement our iterative abstraction method in Python, and tailor the model checker PRISM (Kwiatkowska, Norman, and Parker 2011) for i MDPs to compute robust optimal policies. ... We report the performance of our method on: (1) a UAV motion control, (2) a building temperature regulation, and (3) a spacecraft rendezvous problem.
Researcher Affiliation Academia 1 Radboud University, Nijmegen, the Netherlands 2 University of Oxford, Oxford, United Kingdom 3 University of Birmingham, Birmingham, United Kingdom 4 University of Kentucky, Kentucky, USA 5 University of Twente, Enschede, the Netherlands
Pseudocode Yes Algorithm 1: Sampling-based i MDP abstraction.
Open Source Code Yes Our codes are available via https://gitlab.science.ru.nl/tbadings/sample-abstract
Open Datasets No The paper uses generated noise samples from a simulator and discusses problem benchmarks from other papers (e.g., Cauchi and Abate (2018), Vinod, Gleason, and Oishi (2019)), but it does not provide concrete access information (link, DOI, repository, or specific citation for a dataset) for any publicly available or open dataset used in its experiments.
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) for training, validation, or testing.
Hardware Specification Yes all experiments are run on a computer with 32 3.7GHz cores and 64 GB of RAM.
Software Dependencies No The paper states 'We implement our iterative abstraction method in Python, and tailor the model checker PRISM (Kwiatkowska, Norman, and Parker 2011) for i MDPs to compute robust optimal policies,' but it does not provide specific version numbers for Python, PRISM, or any other libraries.
Experiment Setup Yes In all benchmarks, we use Theorem 1 with β = 0.01, and apply the iterative scheme with γ = 2, starting at N = 25, with an upper bound of 12, 800 samples.