Bi3: A Biplatform, Bicultural, Biperson Dataset for Social Robot Navigation
Andrew Stratton, Phani Teja Singamaneni, Pranav Goyal, Rachid Alami, Christoforos Mavrogiannis
AI summary
Problem
Current social robot navigation datasets lack close-quarters interactions, multi-platform diversity, cross-cultural participant pools, and human perception metrics, hindering the development of robust, socially aware navigation models.
Approach
The authors conducted a controlled lab study where two human subjects repeatedly navigated a constrained workspace alongside a robot running five different autonomous algorithms, collecting multimodal sensor data and user impressions across two international research sites.
Key results
- 10.5 hours of multimodal data capturing dense human-robot navigation encounters
- 74 participants from two distinct cultural sites (USA and France)
- Five autonomous navigation controllers evaluated on two different robot platforms
- Quantitative and qualitative human impressions of robot sociability and workload
Why it matters
Enables researchers to train and validate robot navigation models that safely and naturally operate in dense, constrained indoor environments while accounting for cultural and morphological diversity.
Abstract
We contribute Bi3, a dataset of social robot nav- igation among groups of people in a constrained lab space. Compared to prior data collection efforts for social robot navigation, our dataset is unique in that it features: an original experiment design giving rise to close navigation encounters between two humans and a robot; five different navigation algorithms; two different robot platforms; a diverse participant pool of 74 people recruited from two sites in the USA and France; multimodal data streams including 10.5 hours of human and robot ground-truth motion tracks, RGB video, and user impressions over robot performance. Our analysis of the collected dataset through metrics like interaction density and human velocity suggests that Bi3 represents a benchmark of unique diversity and modeling complexity. Bi3 contributes to- wards understanding how humans and robots can productively mesh their activities in constrained environments, and can be a resource for training models of human motion prediction and robot control policies for navigation in densely crowded spaces.