Bayesian Active Edge Evaluation on Expensive Graphs

Authors: Sanjiban Choudhury, Siddhartha Srinivasa, Sebastian Scherer

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We show that our approach outperforms several state-of-the-art algorithms on a spectrum of planning problems for mobile robots, manipulators and autonomous helicopters. and We evaluate our approach on a spectrum of planning problems including a manipulator and a helicopter. and Table 1 shows the normalized evaluation cost of an algorithm
Researcher Affiliation Academia 1 School of Computer Science and Engineering, University of Washington, USA 2 The Robotics Institute, Carnegie Mellon University, USA
Pseudocode Yes To aid in implementation, we provide a pseudo-code for DIRECT in Alg. 1, 2 and 3. and Algorithm 1: DIRECT (Hact, R, X, c), Algorithm 2: DRD (H , R), Algorithm 3: Weight EC (H , R, i), Algorithm 4: Create Decision Tree (Γ, Hact, R, X, c)
Open Source Code Yes Opensource code and details can be found here: https://github.com/ sanjibac/matlab learning collision checking
Open Datasets Yes A dataset of n worlds is sampled from a designed generative model. Each edge is evaluated on each world to create a test outcome matrix X Rn |T |. A library of paths is created along with a binary membership matrix R Rn m encoding the validity of a path on a world. 10% of the data is used for test, remainder for training. Typical values used are n : 1000, m : 500. Details about the dataset generation are described in [Choudhury et al., 2017b]. Opensource code and details can be found here: https://github.com/ sanjibac/matlab learning collision checking
Dataset Splits No The paper states '10% of the data is used for test, remainder for training' but does not mention a separate validation set or split.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory, or processing units) used for running the experiments.
Software Dependencies No The paper mentions that the code is in a 'matlab learning collision checking' repository, implying MATLAB is used, but no specific version numbers for MATLAB or any other software dependencies are provided.
Experiment Setup No The paper mentions parameters like the threshold η and the bias term α (for BISECT) but does not provide specific values used for these or any other hyperparameters or training configurations for the experimental setup.