Domesticating: Do Autonomous Vehicles Dream of Kangaroos?

This essay begins with the recurring problem of the kangaroo in autonomous driving discourse and follows an encounter between two distinct nonhuman autonomies: the sensing and decision-making systems of autonomous mobility, and the situated ecological autonomy of wildlife. As automated vehicles move across rural roads, extractive zones, and habitat edges, animals enter technical discourse as corner cases, long-tail events, anomalies, and scenarios. I argue that domesticating names the visual-computational operation through which animal life, habitat relations, and sudden roadside appearances are reformatted into absorbable contingencies within autonomous mobility.

Drawing on wildlife encounter footage, roadkill records, autonomous-driving corner-case taxonomies, and recent animal-focused datasets, the essay traces how crossings become scenarios, species-specific habits become behavior labels, and habitat edges become driving-relevant ecologies. In doing so, it complicates benevolent engineering accounts of safety, care, and coexistence in autonomous mobility research and adjacent discussions in Animal-Computer Interaction. The essay asks what ecological relations drop from view when machine sensing, scalable learning, and generalized prediction set the terms of encounter. It argues that contemporary care for the wild, in this engineering sense, now travels through systems that sort, rank, and operationalise contingency, while confidence in real-time, vision-based “learning” and generalization displaces slower and situated ways of “understanding” relations among animals, roads, habitats, and technical systems.