Vlbiman plus plus improves one shot bimanual robot manipulation
VLBiMan++: Expanding the Generalization Boundary of Vision-Language Anchored One-Shot Bimanual Manipulation
Robotics
Summary
Teaching robots to use two hands for diverse tasks usually requires lots of demonstrations and retraining. The paper presents VLBiMan++, which lets robots learn from just one human demonstration and then adapt that skill to different tasks, objects, and environments without more training. By breaking tasks into reusable parts and using vision and language clues, the system handles new objects, cluttered scenes, and even different robot setups. It can also keep working well over time despite interruptions or changes. This approach helps robots become more flexible and reliable in complex, real-world bimanual tasks.
What this means in practice
- •For industrial automation engineers: Enable robotic arms to quickly learn and adapt complex two-handed tasks from a single demonstration across varied products and factory layouts.$Commercial implications: Offers adaptable bimanual robot skill learning to manufacturers, reducing setup time and costs for new assembly lines.
- •For warehouse robotics teams: Deploy dual-arm robots capable of handling diverse objects and cluttered environments without retraining, improving order fulfillment flexibility.
Authors
Huayi Zhou, Wei Gao, Yiyang Han, Kui Jia, Hui Huang
Abstract
Generalizable bimanual robotic manipulation requires a reusable task prior that can persist across increasingly diverse tasks, objects, scenes, embodiments, and execution conditions, thus avoiding the prohibitive cost of large-scale teleoperated demonstrations and policy retraining. In this work, we present VLBiMan++, an extended framework that expands the generalization boundary of vision-language anchored one-shot bimanual manipulation. Starting from a single human demonstration, VLBiMan++ performs task-aware decomposition to identify reusable and adaptable skill components, and employs vision-language grounded geometric adaptation to transfer these skills to novel configurations without retraining. Building on this foundation, we systematically extend generalization along five dimensions: task generalization through diverse and long-horizon skill compositions; object generalization across unseen categories, varying geometries, and more complex articulated or deformable objects; scene generalization under clutter, occlusion, and dynamic interference; embodiment generalization across heterogeneous dual-arm robotic platforms; and deployment generalization through prolonged closed-loop execution under repeated external perturbations. To support this broader scope, we further introduce object-state-aware adaptation and lightweight trajectory optimization mechanisms that accommodate changes beyond simple rigid 6-DoF pose variations while preserving reliable bimanual coordination. Extensive real-world experiments demonstrate that VLBiMan++ maintains strong task success and adaptation capability across these increasingly challenging settings. Overall, VLBiMan++ advances one-shot bimanual manipulation from demonstrating isolated transferability toward a more systematic and scalable framework for generalization across tasks, objects, scenes, embodiments, and long-term deployment conditions.