AISL Attendant Behavior Dataset

The AISL Attendant Behavior is a collection of behavior of professional caregivers attending elderly persons with dementia. This datasets consists of behavior sequences of caregivers and elderly persons (position and velocity) during attendance. The dataset was recorded in indoor and outdoor environments of Swarabikai Fukushimura hospital (link).

Description

People bahavior measurement and analysis are important to design systems which interact with with people. In particular, for attendant service robots, it is crucial to understand how humans attend another person to achieve natual and comfortable services, and it requires analyzing people behavior in diverse situations. However, to our knowledge, there is no dataset which provides people behavior in attendance. Due to this reason, we decided to develop a system which enables long-term and wide-area people behavior measurement and provide a dataset which consists of real professional human's attendant behavior data.

Data Acquisition

The dataset was recorded using a portable people tracking system equipped with a 3D LIDAR. In this system, an observer carries the system and follows the persons to be measured while keeping them in the sensor view. The system simultaneously estimates the sensor pose and tracks the target persons (see Fig. 1, and the detail is described in the related paper). We recorded behavior of professional caregivers and elderly persons for one month, and with this dataset, we provide the recorded their behavior data (position and velocity sequences). The durations of indoor and outdoor sequences are about 2500[sec] and 1000[sec], respectively.

Fig. 1. The portable people behavior measurement system. An observer carries the system equipped with a 3D LIDAR and follows people to be measured while keeping them in the sensor view.
Fig. 2 (a). A snapshot of the recording. A professional caregiver attends an elderly with dementia. An observer equipping the system follows them to measure their behavior.
Fig. 2 (b). An example of recorded range data. For privacy reasons, this dataset provides only behavior data extracted from the range data sequences.
Contents

The dataset consits of two directories ("trajectories_indoor" and "trajectories_outdoor"), and each directory contains a set of csv files which consist of sequences of:

  1. Time [msec]
  2. Caregiver's position [m]
  3. Caregiver's velocity [m/s]
  4. Elderly's position [m]
  5. Elderly's velocity [m/s]
Note that while the position and velocity data in indoor environments are 2D, ones in outdoor environments are 3D since the outdoor environment had large undulating. You can directly read the csv files using numpy.loadtxt function. The dataset also constains an example program (plot.py) to plot the distributions of the caregivers' relative position with respect to the elderly person (see Fig. 3).
Fig. 3 (a). The distribution of the caregivers' position in an indoor environment. The orange triangle indicates the elderly's walking direction.
Fig. 3 (b). The distribution of the caregivers' position in an outdoor environment.

Measurement System

We also provide the software of the system as open source ROS packages. You can find them at the following links;

  1. hdl_graph_slam
  2. hdl_localization
  3. hdl_people_tracking (will be available soon)

Acknowledgement

We would like to thank O. Kohashi, S. Yamamoto, and T. Gomyo for allowing us to conduct the field test in Sawarabikai Fukushimura hospital and their excellent cooperation during the test.

Agreement

The AISL attendant behavior dataset is now made available for research purpose only.
The researcher(s) is free to use the AISL attendant behavior dataset by obeying and agreeing the following restrictions on the dataset:

  1. The dataset will not be further distributed, published, copied, or further disseminated in any way whether for profit or not.
  2. All the data will be used for the purpose of scientific researches only. The AISL attendant behavior dataset, in whole or in part, will not be used for any commercial purpose in any form.
  3. All technical papers, documents and reports which use the AISL attendant behavior dataset will acknowledge the use of the dataset by a citation to
    4. “Kenji Koide, Jun Miura, and Emanuele Menegatti, A Portable 3D LIDAR-based System for Long-term and Wide-area People Behavior Measurement, IEEE Transactions on Human-Machine Systems, 2017, (under review)”.

Please fill in all fields and check the box below to continue download the dataset.

Name
Institution   
Email

  By checking this check box, it means that you agree to our restrictions.