SFU-store-nav: A multimodal dataset for indoor human navigation.

Area of Science:

• Robotics
• Human-Robot Interaction
• Machine Learning
• Computer Vision

Background:

• Autonomous robot navigation requires understanding human behavior and intent.
• Simulating real-world scenarios is crucial for collecting relevant human-robot interaction data.

Purpose of the Study:

• To create a dataset of human gestures, movements, and behaviors indicative of navigational intent.
• To support the development of autonomous robot navigation systems.

Main Methods:

• Conducted experiments in a robotics lab simulating a shopping scenario.
• Collected visual and motion capture data from 108 human participants interacting with a Pepper robot.

Main Results:

• Acquired comprehensive visual recordings of human-robot interactions.
• Obtained precise motion capture data detailing human participant positions and orientations.

Conclusions:

• The dataset provides valuable insights into human navigational intent for robot navigation.
• This resource is beneficial for researchers in robotics, machine learning, and computer vision.