Search papers, labs, and topics across Lattice.
This paper introduces a novel approach to robotic grasp synthesis using a Lie Group-constrained MeanFlow method, which significantly enhances the speed of generative grasping. By coupling algebraic semigroup consistency with Riemannian Conditional Flow Matching on the product Lie group, the proposed method achieves reliable grasp sampling in under five network evaluations. The results show that this approach not only matches the performance of leading generative models but also operates at millisecond-scale latency, enabling effective real-world application without the need for additional training.
Achieving grasp synthesis in under five evaluations, this method accelerates generative models by up to 39 times while maintaining state-of-the-art performance.
Grasp synthesis is a core task in robotic manipulation, for which the solution typically forms a multimodal distribution rather than a point estimate. Generative robotic grasping aims to learn this distribution with deep generative models such as diffusion and flow-based approaches. The iterative nature of such generative models makes them flexible and generalizable; however, multi-step sampling impedes the time-critical operation required in robotics. We devise an approach to fast generative grasping based on MeanFlow on the product Lie group $\mathcal{G} = \mathrm{SO}(3) \times \mathbb{R}^3$. The training objective couples a purely algebraic semigroup consistency condition with Riemannian Conditional Flow Matching on $\mathcal{G}$ that anchors the average velocity to the data distribution. The resulting Lie Group-constrained MeanFlow formulation samples reliable grasps in $\leq 5$ network evaluations, matching the grasp generation performance of state-of-the-art diffusion and flow-based models on the ACRONYM dataset at millisecond-scale inference latency (up to $39\times$ speed-up). We further demonstrate that the approach directly translates to real-world robotic grasping without additional training or domain adaptation, exhibiting robust grasp synthesis under observation noise.
