arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2501.18229cs.RO
收藏

GPD:用于运动规划的引导式多项式扩散

GPD: Guided Polynomial Diffusion for Motion Planning

Robotic Research Center, IIIT Hyderabad(印度国际信息技术研究院海得拉巴分校机器人研究中心);Carnegie Mellon University(卡内基梅隆大学);TCS Research(塔塔咨询服务公司研究院);University of Tartu(塔尔图大学)

Ajit Srikanth, Parth Mahanjan, Kallol Saha, Vishal Mandadi, Pranjal Paul, Pawan Wadhwani, Brojeshwar Bhowmick, Arun Singh, Madhava Krishna

AI总结:

本文提出在轨迹参数空间(表示为Bernstein系数)中进行扩散的GPD方法,结合新颖的拼接算法,显著提高了代价函数引导的有效性与推理速度,仅用单个引导模型即可生成无碰撞轨迹,性能优于现有SOTA方法。

摘要:

基于扩散的运动规划器正变得流行,这得益于其成熟的性能提升,该提升源于样本的多样性以及在推理过程中直接引入新约束的便利性。然而,扩散过程的一个主要局限是需要大量的去噪步骤,特别是当去噪过程与基于梯度的引导相结合时。在本文中,我们引入了在轨迹的参数空间中进行扩散的方法,其中参数表示为Bernstein系数。我们表明,这种表示方法极大地提高了代价函数引导的有效性和推理速度。我们还引入了一种新颖的拼接算法,该算法利用扩散生成的轨迹的多样性,仅使用单个代价函数引导模型即可生成无碰撞轨迹。我们证明了我们的方法优于当前用于机械臂的SOTA基于扩散的运动规划器,并对关键组件进行了消融研究。

英文摘要:

Diffusion-based motion planners are becoming popular due to their well-established performance improvements, stemming from sample diversity and the ease of incorporating new constraints directly during inference. However, a primary limitation of the diffusion process is the requirement for a substantial number of denoising steps, especially when the denoising process is coupled with gradient-based guidance. In this paper, we introduce, diffusion in the parametric space of trajectories, where the parameters are represented as Bernstein coefficients. We show that this representation greatly improves the effectiveness of the cost function guidance and the inference speed. We also introduce a novel stitching algorithm that leverages the diversity in diffusion-generated trajectories to produce collision-free trajectories with just a single cost function-guided model. We demonstrate that our approaches outperform current SOTA diffusion-based motion planners for manipulators and provide an ablation study on key components.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑