search

DARoSLab / cear_dataset_utils

Utilities for CEAR dataset: https://daroslab.github.io/cear/
2 stars 0 forks source link

Time offsets in the dataset #1

Open FORREST1901 opened 4 months ago

FORREST1901 commented 4 months ago

Your dataset offers the first comprehensive multi-sensor data for an Agile Quadruped Robot, which I find truly impressive as it fills a significant gap. Thank you for your valuable contribution to the community.

I am interested in using your dataset and citing your work, but I have a few questions regarding its application:

  1. The paper notes that synchronization was achieved using the robot's motion patterns instead of hardware synchronization. I'd like to know if the time offset between the various sensors have been adjusted in the data, or if they remain uncorrected. Where can I find details about these time offsets? And among different dataset, the time offset is varies or consistent?
  2. The code on GitHub for merging data into the ROSbag format does not seem to process event data, as that part is commented out.
  3. The GitHub code is written in Python2. Could you indicate if many modifications are necessary to make it compatible with a Python3 environment?
ShifanZhu commented 4 months ago

Hi, thank you for your interest in our work!

  1. Synchronization: All ground-truth poses have been synchronized with the event camera's clock source. Other sensors' data is not synchronized, you can find the time offset parameters in the Calibration section.

  2. Data Conversion: To convert AEDAT4 event data to ROS bag format, you can use this algorithm.

  3. Compatibility: The code has been tested on Ubuntu 18.04 and ROS Melodic, which uses Python 2.7 by default. Minimal modifications should be necessary to adapt it for a Python 3 environment.