GestureLint

GestureLint is a web + desktop app for teams building social/service robots that need their robot’s gestures, gaze, and proxemics to be understandable and culturally safe. You import ROS bag/video logs (or live stream), mark intended meanings (e.g., “follow me,” “stop,” “apology”), and the app automatically flags likely misinterpretations: ambiguous hand trajectories, conflicting gaze vs motion, too-fast timing, unsafe interpersonal distance, and repeated “dead time” that reads as broken. It generates a short “interaction lint report” with clips, severity, and concrete fixes (timing curves, pose constraints, alternative gesture templates). It also supports quick remote tests: send a clip to a panel, collect forced-choice meaning + confidence, and compare against intent. This is not magic: it won’t design perfect HRI, but it will reduce preventable confusion and speed iteration for small teams without dedicated HRI researchers.