This work highlights the application of motion-based, socially-aware robotic systems that are designed to engage in close-proximity, collaborative interactions with humans. By programming normative pedestrian behaviors - things like turn-taking, yielding, and spatial negotiation - it is possible to facilitate interactions that are both intuitive and safer for human collaborators and bystanders. Social intelligence datasets are a necessary component of intuitive robot design and extend beyond conventional objective-function optimization.

Download the full article here.