Layered Integration of Visual Foundation Models for Enhanced Robot Manipulation and Motion Planning

EasyChair Preprint 13187

Abstract

Robotics research has seen significant advancements in recent years, particularly in the realms of visual perception, manipulation, and motion planning. This paper proposes a novel approach termed Layered Integration of Visual Foundation Models (LIVFM) aimed at enhancing robot manipulation and motion planning tasks. LIVFM integrates multiple visual perception models in a layered fashion, leveraging the strengths of each model to overcome their individual limitations. By combining the outputs of these models, robots can achieve enhanced understanding of their environment, leading to improved manipulation capabilities and more robust motion planning strategies. This paper presents the theoretical framework of LIVFM, discusses its implementation details, and provides experimental results demonstrating its effectiveness in various robotic scenarios.

Keyphrases: Integration, Layered models, Robotics, enhanced performance, manipulation, motion planning, visual perception

@booklet{EasyChair:13187,
  author    = {Anthony Lambert and Wahaj Ahmed},
  title     = {Layered Integration of Visual Foundation Models for Enhanced Robot Manipulation and Motion Planning},
  howpublished = {EasyChair Preprint 13187},
  year      = {EasyChair, 2024}}