Object Detection for Embodied Agents using Sensory Commutativity of Action Sequences

Hugo Caselles-Dupré (1, 2, 3) , Michael Garcia Ortiz (4) , David Filliat (2, 1)

1 U2IS - Unité d'Informatique et d'Ingénierie des Systèmes
2 Flowers - Flowing Epigenetic Robots and Systems
3 SoftBank Robotics Europe
4 City University London

We study perception in the scenario of an embodied agent equipped with first-person sensors and a continuous motor space with multiple degrees of freedom.We introduce a method for movable and immovable object detection that exploitsthe commutative properties of action sequences. Our method is based on playingan action sequence in two different orders from the same starting point, and testedon 3D realistic robotics setups.

Domaines: Robotique [cs.RO]

Hugo Caselles-Dupré, Michael Garcia Ortiz, David Filliat. Object Detection for Embodied Agents using Sensory Commutativity of Action Sequences. NeurIPS 2020 Workshop on BabyMind, Dec 2020, Vancouver / Virtual, Canada. ⟨hal-03123933⟩