Iterative Continuous Convolution for 3D Template Matching and Global Localization

Authors: Vitor Guizilini, Fabio Ramos

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results show substantial speed gains over standard discrete convolution techniques, such as sliding window and fast Fourier transform, along with a significant decrease in memory requirements, without accuracy loss.
Researcher Affiliation Academia Vitor Guizilini, Fabio Ramos School of Information Technologies, The University of Sydney, Australia e-mail: {vitor.guizilini@;fabio.ramos}sydney.edu.au
Pseudocode Yes Algorithm 1 Pseudo-code for the continuous convolution between two Hilbert Maps
Open Source Code No The paper does not contain an unambiguous statement of code release or a link to a repository for the methodology described.
Open Datasets No Two large-scale real datasets were used, representing a structured corridor (n = 474557) and an outdoor area (n = 670584).
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning.
Hardware Specification Yes All computations were performed on a i7/2.60 x 8 GHz notebook, with multi-threading wherever possible.
Software Dependencies No The paper mentions the 'PCL library' but does not provide specific version numbers for any software dependencies required to replicate the experiments.
Experiment Setup Yes To enforce sparsity, only the Q nearest clusters from each data point are selected to produce its feature vector (in all experiments, a value of Q = 3 was used).