User-Friendly Enhancements to the Scout Mini Robot for Collecting Natural Human-Robot Interaction Data

Abstract

Collecting data for social robot navigation often involves strictly controlled scenarios over limited time frames, making it impractical to scale for real-world deployments. Consequently, objects such as people and furniture are often staged, limiting the realism of collected data. To address this challenge, we built an autonomous indoor mobile robot capable of navigating the Fuse building and delivering objects, while collecting natural human-robot interaction data. We incorporated a seven-inch touchscreen into a customized Agile X Scout Mini platform, providing a user-friendly interface through an onboard computer. To increase accessibility and durability, we developed a rotatable touchscreen mount to protect the display from impacts while allowing adjustable viewing angles. Furthermore, we replaced Orin AGX’s original wireless card with an Intel AX210 and mounted external antennas to provide a reliable wireless network connection. We also developed a robot dashboard to display sensor status and incorporated a resource-aware monitoring mechanism for troubleshooting. Since dashboard topic rates differed from sensor topic publication rates, we implemented a node that aggregates and publishes sensor topics as a status message, allowing for more accurate monitoring. These contributions enhance user accessibility, improve development and troubleshooting processes, and provide protection for hardware components to support continuous deployment for collecting natural human-robot interaction data.