Mobile robots carrying food in restaurants are here. What service behavior norms do people expect them to follow? This paper evaluates robot waiter algorithms and service parameters for a robot serving two participants at a simulated cocktail event, varying body-storming inspired context variables such as: "hunger level" and "relationship to each other," robot delivery algorithms (lead, follow, ambient), and participant pose (standing, seated). In the within-subjects design, pairs of people were given a series of context prompts, and told to participate as felt natural. Output variables included whether they took food and post-trial survey ratings of the robot. The results show a positive correlation between food taking (or feelings of obligation to take food) and human OR robot initiative, relative to a mixed-ambient algorithm with no explicit leader. The robot waiter that initiates is the clearest and most noticeable. There were also some challenges: people in conversation would sometimes forget or delay calls for cupcakes, ambient robot motion was hardest to notice, and bringing food one person ordered to the other was unforgivable. When in doubt, go to the middle. Finally, participants enjoyed the robot spinning, describing it as a dessert tray which attracted their eyes to the robot.

Spatial formations can give many social cues, such as illustrating a group of people are having a conversation (social affiliation), or that they are trying to move swiftly through a space (functional goal). This work explored how people perceive varied robots formations while navigating through a space and approaching people. Evaluation occurred across four different geometric formations: wedge, v-shape, vertical line, and horizontal line (Fig 3). Two studies were conducted: the first being an exploratory study of three robots navigating through a public space, and the second being a controlled user study of the same robots approaching humans in different formations. Results showed that triangle shapes were generally received more positively than lines, with wedge being the viewed as harmless, polite, welcoming, and encouraging the human to join the robot group, whereas horizontal line was seen as threatening and unwelcoming. From a path planning perspective, v-shape and wedge were also more robust to controller variance. Results from this work show that formation impacts how people perceive robots, and as a result may impact task success. Future researchers can use these results to inform their behavior design for multi-robot groups to increase task success and desired communication effects.

Our neural architecture search algorithm progressively searches a tree of neural network architectures. Child nodes are created by inserting new layers determined by a transition graph into a parent network up to a maximum depth and pruned when performance is worse than its parent. This increases efficiency but makes the algorithm greedy. Simpler networks are successfully found before more complex ones that can achieve benchmark performance similar to other top-performing networks.