Social Robot Navigation – Opportunities, Algorithms, Tools, and Systems

Mobile robots have recently become more prevalent in human environments, including delivery robots on sidewalks, guide robots in museums and airports, and cleaning robots at shopping malls. However, human acceptance of these robots is limited because the cutting-edge robots have yet to demonstrate understanding of the social rules followed by humans while navigating. As a result, these robots typically exhibit behavior that is inefficient (e.g., pausing extensively) or inappropriate (e.g., intruding into personal spaces); these behaviors are especially noticeable in densely-populated settings. Existing research has shown that simply treating humans as dynamic obstacles is insufficient. It has also been suggested that addressing human behavior uncertainty, modeling multi-agent interaction, and factoring diverse environmental contexts are key challenges to be addressed. Much is unknown about each of the key challenges and, with the rising popularity of mobile robots and autonomous vehicles, it is increasingly urgent to develop new solutions in the field of social navigation.

This Research Topic seeks to address the challenges of social navigation to bring robots closer to seamless integration into human environments. It can be broken down into four themes: opportunities, algorithms, tools, and systems. “Opportunities” refers to principled knowledge about how humans behave while navigating, such as respecting personal space, communicating navigation intentions, and staying in group formations. They are identified via human-robot interaction research, often through user studies. “Algorithms” refers to methodological advances in modeling human behavior or robot navigation, such as detecting and predicting future human actions, learning through deployment in human environments, and systematic modeling of the opportunities identified above. “Tools” refers to simulators, datasets, and benchmarks that aid in the training and evaluation of social navigation models. These tools provide valuable data for producing credible models and equal grounds to compare algorithm performances. “Systems” refers to the finalized integration into a robot and deployment of the robot in the real world. Through this, “opportunities” can be refined to better reflect social navigation requirements, “algorithms” can be validated on their feasibility and generalizability, and “tools” can be evaluated on their sim-to-real transferability.

We aim to compile state-of-the-art research in social navigation, involving robotic agents that are situated in complex human environments. We invite interested researchers to submit original research or works with significant additional contributions to prior work. Manuscripts may focus on theories, technical contributions, user studies, systems, case studies, or surveys of existing work.

The themes include, but are not limited to:
● Agent intent prediction, activity recognition, behavior understanding
● Multi-agent coordination and communication
● Modeling, simulating, or factorizing complex environments
● Human-aware path-planning, navigation, and control
● Uncertainty modeling, safe learning, and robustness
● Benchmarks, datasets, metrics, and evaluations in real-world environments
● Multimodal, multisensory, or multiview representation learning
● Scene characterization; scenario generation
● Hybrid learning-based and model-based frameworks

Keywords: Social navigation, Human-aware navigation, Socially aware robot navigation, Social robotics, Human-robot interaction, Path planning

Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.