Safe Interval Path Planning with Kinodynamic Constraints

Authors: Zain Alabedeen Ali, Konstantin Yakovlev

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In the experimental evaluation, we show that the key property of the original SIPP still holds for the modified version: it performs much fewer expansions compared to A* and, as a result, is notably faster. To empirically evaluate it, we have conducted a wide range of experiments in which we compare SIPP-IP to several baselines that include other (noncomplete) variants of SIPP one may think of and A*.
Researcher Affiliation Collaboration Zain Alabedeen Ali1, Konstatin Yakovlev2, 3 1 Moscow Institute of Physics and Technology, Moscow, Russia 2 Federal Research Center for Computer Science and Control of Russian Academy of Sciences, Moscow, Russia 3 AIRI, Moscow, Russia
Pseudocode Yes The pseudo-code of SIPP-IP is presented in Algorithm 1.
Open Source Code Yes The C++ source code of all planners is publicly available https://github.com/PathPlanning/SIPP-IP.
Open Datasets Yes We have used five different maps from the Moving AI benchmark (Sturtevant 2012) for the experiments: empty (sized 64x64), room (64x64), warehouse (84x170), random (128x128) and Sydney (256x256).
Dataset Splits No The paper uses "generated instances" and "successfully solved instances" but does not explicitly state dataset splits for training, validation, and testing of models, only the test scenarios setup.
Hardware Specification Yes The experiments were conducted on a PC with Intel Core i7-10700F CPU @ 2.90GHz 16 and 32Gb of RAM.
Software Dependencies No The paper mentions that the source code is C++ but does not provide specific version numbers for compilers, libraries, or other software dependencies.
Experiment Setup Yes The agent was modeled as a disk whose diameter equals the length of the grid cell. The configuration is defined as (x, y, θ, vel) where x, y are the coordinates of the cell, θ {0 , 90 , 180 , 270 } is the orientation, and vel {0, 2} is the velocity of the agent. ... The time step is chosen to be 0.1 s. Collision checking is conservative. ... We imposed a limit of 100, 000, 000 of generated nodes for all algorithms.