aiqbench analysis
This research introduces BiCompoDiff, a novel framework combining diffusion models with energy-based optimization to solve complex bimanual tasks like object reorientation and handover. By integrating gradient guidance during reverse diffusion, the method effectively balances multiple constraints such as collision avoidance and trajectory smoothness while demonstrating strong sim-to-real transfer capabilities. The approach represents significant progress in advanced robotic manipulation through sophisticated generative modeling techniques applied directly to physical control problems.
Grasp, Handover, Rotate: Bimanual Object Reorientation via Compositional Diffusion and Energy-Based Optimization
arXiv:2607.21341v1 Announce Type: new Abstract: Bimanual object reorientation - picking an object, handing it over between two arms, and placing it in a desired target pose - is valuable when direct placement from the initial grasp is infeasible due to collisions, kinematic constraints, or poor final orientation. However, achieving this under multiple competing objectives remains challenging. We introduce BiCompoDiff, a compositional diffusion and energy-based framework that jointly optimizes grasp selection, handover, regrasp, and motion planning under multiple constraints. By combining a pretrained grasp diffusion model with bimanual planning energy-based models (EBMs), our method injects gradient guidance during reverse diffusion to enforce collision avoidance, trajectory smoothness (via differentiable inverse kinematics), handover feasibility, and regrasp safety. Annealed MCMC sampling further refines grasp poses over the composite energy landscape. Experiments across diverse simulated household reorientation tasks demonstrate that BiCompoDiff achieves over 20% higher success rates and up to 37% smoother trajectories (measured by joint displacement) compared to strong sampling-based baselines. Real-world validation confirms effective sim-to-real transfer and robust performance on challenging scenes.
Abstract:Bimanual object reorientation - picking an object, handing it over between two arms, and placing it in a desired target pose - is valuable when direct placement from the initial grasp is infeasible due to collisions, kinematic constraints, or poor final orientation. However, achieving this under multiple competing objectives remains challenging. We introduce BiCompoDiff, a compositional diffusion and energy-based framework that jointly optimizes grasp selection, handover, regrasp, and motion planning under multiple constraints. By combining a pretrained grasp diffusion model with bimanual planning energy-based models (EBMs), our method injects gradient guidance during reverse diffusion to enforce collision avoidance, trajectory smoothness (via differentiable inverse kinematics), handover feasibility, and regrasp safety. Annealed MCMC sampling further refines grasp poses over the composite energy landscape. Experiments across diverse simulated household reorientation tasks demonstrate that BiCompoDiff achieves over 20% higher success rates and up to 37% smoother trajectories (measured by joint displacement) compared to strong sampling-based baselines. Real-world validation confirms effective sim-to-real transfer and robust performance on challenging scenes.
Robotics (cs.RO)
arXiv:2607.21341 [cs.RO]
(or arXiv:2607.21341v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2607.21341 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 257 words
Related coverage
The Machines that Make the Machines
NVIDIA · Oct 7, 2026
LeRobot v0.6.0: Imagine, Evaluate, Improve
Hugging Face · Jul 7, 2026
EquiDexFlow: Contact-Grounded SE(3)-Equivariant Dexterous Grasp Generative Flows
arXiv · Jun 16, 2026
EquiDexFlow: Contact-Grounded SE(3)-Equivariant Dexterous Grasp Generative Flows
arXiv · Jun 12, 2026