kinova-gen3-0eba3081·1 events·first seen Aliases: Kinova Gen3
Researchers introduce FurnitureVLA, a Vision-Language-Action model system for real-scale bimanual robot furniture assembly, addressing tasks with up to 7 subtasks and 1550 control steps. The approach combines a progress-enhanced VLA finetuned on semantically grounded subtasks with joint action and continuous progress signal prediction to enable automatic subtask transitions and reduce compounding errors. The system improves simulation success from 48% to 80% over baselines across three furniture types and validates on a real Kinova Gen3 robot platform. This represents the first systematic study of real-scale bimanual furniture assembly using VLAs.