CTO-SLAM: Contour Tracking for Object-Level Robust 4D SLAM

Authors: Xiaohan Li, Dong Liu, Jun Wu

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The CTO-SLAM system is verified on both KITTI and VKITTI datasets. The experimental results demonstrate that our system effectively addresses cumulative errors in long-term spatiotemporal association and hence obtains substantial improvements over the state-of-the-art systems.
Researcher Affiliation Academia Xiaohan Li1, Dong Liu1, Jun Wu2* 1Institute of Advanced Technology, University of Science and Technology of China 2Fudan University
Pseudocode No The paper includes diagrams illustrating system pipelines (Fig. 1, Fig. 2, Fig. 3) but does not contain explicit pseudocode or algorithm blocks.
Open Source Code Yes The source code is available at https://github.com/real Xiaohan/CTO-SLAM.
Open Datasets Yes The CTO-SLAM system is verified on both KITTI and VKITTI datasets. The KITTI Tracking dataset serves as a valuable resource for evaluating CTO-SLAM. The VKITTI dataset is derived from the KITTI tracking benchmark.
Dataset Splits No The paper mentions using KITTI and VKITTI datasets for evaluation but does not specify how the data was split into training, validation, and test sets with specific percentages or counts.
Hardware Specification No The paper does not explicitly describe the specific hardware (e.g., GPU models, CPU types, memory) used to run the experiments.
Software Dependencies No The paper mentions using 'DROID-SLAM' and 'ORB-SLAM II' as components or baselines but does not provide specific version numbers for these or other software dependencies.
Experiment Setup Yes Dynamic object pose initialization is crucial for accurately positioning objects in the global map when they are first observed. ... it is reasonable to assign pitch and roll angle to zero. ... we uniformly extract ten contour keypoints for an object and make a robust guess with RANSAC. ... the object motion initialization algorithm is periodically called to correct and align the ground over every 5 keyframes. ... we increase the weight of sampling-based poses in factor graph for better optimization.