Safe Exploration in Finite Markov Decision Processes with Gaussian Processes

Authors: Matteo Turchetta, Felix Berkenkamp, Andreas Krause

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We demonstrate our method on digital terrain models for the task of exploring an unknown map with a rover. In this section, we demonstrate Algorithm 1 on an exploration task. We consider the setting in [14], the exploration of the surface of Mars with a rover.
Researcher Affiliation Academia Matteo Turchetta ETH Zurich matteotu@ethz.ch Felix Berkenkamp ETH Zurich befelix@ethz.ch Andreas Krause ETH Zurich krausea@ethz.ch
Pseudocode Yes Algorithm 1 Safe exploration in MDPs (Safe MDP)
Open Source Code Yes The code for the experiments is available at http://github.com/befelix/Safe MDP.
Open Datasets Yes In our experiments we use digital terrain models of the surface of Mars from the High Resolution Imaging Science Experiment (Hi RISE), which have a resolution of one meter [12].
Dataset Splits No The paper describes an exploration algorithm and its evaluation, but does not provide specific training/validation/test dataset splits for reproducibility in the traditional sense of supervised learning.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., CPU, GPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using a 'GP framework' and 'Matérn kernel' but does not specify particular software libraries or their version numbers (e.g., Python, PyTorch, scikit-learn, etc.) that would be needed for replication.
Experiment Setup Yes The lengthscales are set to 14.5 m and the prior standard deviation of heights is 10 m. We assume a noise standard deviation of 0.075 m. we fix βt to a constant value, βt = 2, 8t 0.