Sensor-based Pallet Activity Recognition in Logistics (SPARL Version 2) - A multi-modal Dataset

SPARL is a freely accessible data set for sensor-based activity recognition of pallets in logistics. The data set consists of 20 recordings from three scenarios. A description of the scenarios can be found in the protocol file.

Four different sensors were used simultaneously for all recordings:

• MSR Electronics MSR 145
  • Sampling rate 50 Hz
• MBIENTLAB MetaMotionS
  • Sampling rate 100 Hz
• Kistler KiDaQ Module 5512A
  • Sampling rate 100 kHz
  • the raw data is also downsampled to 5 kHz and 20 kHz for easier processing 
• Holybro Flightcontroller PX4FMU
  • The board uses two accelerometers and two gyroscopes, all with a sampling rate of 1000 Hz
    • Accelerometer 1: IvenSense MPU6000 
    • Accelerometer 2: STMicroelectronics LSM303D 
    • Gyroscope 1: IvenSense MPU6000
    • Gyroscope 2: STMicroelectronics L3GD20

The recordings were accompanied by three logitech Mevo Start cameras, of which all recordings are included anonymously in the data set. 

The videos were annotated by one person in each frame. For this purpose, the annotation tool SARA was used, which can be found here. The JSON schema used for annotation is also included in the SPARL dataset. The R code used our evaluation can be found in GitHub.

If you have any questions about the dataset, please contact: sven.franke@tu-dortmund.de

If you use this dataset for research, please cite the following paper: “Data-driven, sensor-based taxonomy for environmental life cycle assessment of pallets”, Nr. 20 (2024): Logistics Journal: Proceedings, DOI: 10.2195/lj_proc_franke_en_202410_01