VendoBot: A Projection-Augmented Delivery Robot for Improving Intent and Motion Legibility

Описание: This video presents VendoBot, a projection-augmented autonomous delivery robot designed to improve how humans understand and anticipate robot motion in shared indoor environments. VendoBot communicates its navigation intent by projecting visual cues directly onto the floor in front of the robot during navigation.

The system integrates a ROS-based navigation stack with a projector-equipped Raspberry Pi 5, a PostgreSQL-backed order management pipeline, and Android applications for both proximate and remote interaction. During motion, the projection engine subscribes to the robot’s local planner and renders either simple directional arrows or a continuous animated visualization of the robot’s short-horizon planned trajectory in real time.

The video demonstrates the full end-to-end workflow, including order placement, autonomous navigation, and intent projection during hallway encounters. It also illustrates how different projection styles influence bystander interpretation, spatial coordination, and perceived motion predictability.

This work was evaluated in an exploratory deployment study involving both delivery recipients and bystanders in a semi-public university hallway. Results indicate that trajectory-based projection improves perceived intent legibility, predictability, and user comfort compared to arrow-based or no-projection conditions.

The paper is presented in the 20th Annual IEEE/ACM International Conference on Human-Robot Interaction (HRI 2026), Edinburgh, Scotland, UK.

Source code and system implementation:
https://github.com/hshaon/vendobots

Includes the projection engine, base-station interface, Android apps, and deployment utilities.