SCC-IRG: Public Space Robotics: Community-Driven Models for Social Navigation and Communication

This Smart and Connected Communities Integrative Research Grant (SCC-IRG) project supports research that aims to develop systems that help robots navigate public areas in a safe and socially acceptable manner. While robots have great potential to improve communities by providing services such as delivery, safety patrol, and sanitation, they are still limited in their ability to autonomously manage the complexities of real-world public walkways, which often feature sprawling obstacles, unmaintained sidewalks, and varying weather conditions. Robots may also find it challenging to navigate among people who are on their phones, not paying attention, using a cane or a walker, pushing a stroller, or walking their dog. This can lead to unsafe interactions and hinder the public acceptance necessary to realize the full potential of robotic services. However, robots cannot simply follow hard traffic rules and must learn to adapt and adjust their behavior based on the scenarios they encounter. This project will engage directly with community members to collect data and train new algorithms on how people want robots to behave in public. Intended outputs of this research are improved ways for robots to navigate in public and communicate with community members. By developing socially appropriate navigation and communication for these robots, this research seeks to create safer and more effective robotic systems. This work serves the national interest by advancing scientific progress in human-robot interaction, enhancing public welfare and safety on community walkways, and supporting the nation's leadership in developing and deploying civic robotic technology with direct citizen engagement, thereby enhancing the economic potential of public robots and improving community quality of life.

Developing safe and socially appropriate behaviors for public-area robots requires intelligent methods for integrating human feedback from realistic scenarios into the robot's learning process. It also requires new design patterns that enable robots and people to communicate effectively when on the sidewalk. This project aims to enhance robot social navigation and communication on public walkways through continuous community engagement, including co-designing interaction scenarios, refining pedestrian models and sidewalk simulations, and training robot navigation and communication systems. The project will employ a phased development process that combines 3D simulation environments, laboratory studies, and field tests. A key technical approach involves using a novel machine-learning framework where community members will participate in pilot demonstrations in both virtual and physical settings to train the robot's navigation and communication systems. This method seeks to enable efficient training of socially aware behaviors and design patterns that can be transferred across different types of robots. The performance of the resulting robotic systems will be rigorously evaluated through laboratory studies and field tests within the community, using both wheeled and legged robots to test the generalizability of the learned behaviors. The Lawrenceville Corporation, a local community organization in Pittsburgh, and Carnegie Mellon's Metro21 Smart Cities Institute will help facilitate engagement with community members, grounding the project in real-world complexities and connecting it to the everyday experiences of residents. The project will also partner with the Urban Robotics Foundation, which will contribute its extensive knowledge of urban robot systems and support the research team in translating findings into policy and technical standards recommendations.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.