Sim-to-real transfer via a Style-Identified Cycle Consistent Generative Adversarial Network: Zero-shot deployment on robotic manipulators through visual domain adaptation

Lucía Güitta-López, Lionel Güitta-López, Jaime Boal, Alvaro Jesús López López

Abstract: Highlights•Our zero-shot method efficiently bridges the reality gap in robotic manipulators.•The DRL agent can generalize its knowledge to address different real objects.•Our original SICGAN efficiently translates images from sim-to-real and real-to-sim.

External IDs:dblp:journals/eaai/GuittaLopezGBL25