Scholarlink

This chapter explores how large language models (LLMs) enable robotic planning, control, and mapping through techniques like supervised fine-tuning (SFT) and direct preference optimization (DPO). It covers transformer-based models such as SayCan and RT-1, which facilitate multimodal task execution, and diffusion-based policies for flexible action generation. The chapter also highlights security risks in AI-driven robots and emphasizes the need for robust safety measures, anomaly detection, and interpretability to ensure safe deployment.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Foundation Models in Robotics

  • 章节

  • 发表日期：

    2025年

查看全文
  • Alishba Imran,
  • Keerthana Gopalakrishnan

摘要

This chapter explores how large language models (LLMs) enable robotic planning, control, and mapping through techniques like supervised fine-tuning (SFT) and direct preference optimization (DPO). It covers transformer-based models such as SayCan and RT-1, which facilitate multimodal task execution, and diffusion-based policies for flexible action generation. The chapter also highlights security risks in AI-driven robots and emphasizes the need for robust safety measures, anomaly detection, and interpretability to ensure safe deployment.