Embodied AI in Machine Learning -- is it Really Embodied?
This is an incremental critique for researchers in robotics and AI, highlighting limitations in achieving true embodiment.
The paper critiques current Embodied AI approaches, arguing that AI-powered robots are only weakly embodied and inherit issues from traditional AI, while reviewing cross-embodiment learning and identifying roadblocks for progress.
Embodied Artificial Intelligence (Embodied AI) is gaining momentum in the machine learning communities with the goal of leveraging current progress in AI (deep learning, transformers, large language and visual-language models) to empower robots. In this chapter we put this work in the context of "Good Old-Fashioned Artificial Intelligence" (GOFAI) (Haugeland, 1989) and the behavior-based or embodied alternatives (R. A. Brooks 1991; Pfeifer and Scheier 2001). We claim that the AI-powered robots are only weakly embodied and inherit some of the problems of GOFAI. Moreover, we review and critically discuss the possibility of cross-embodiment learning (Padalkar et al. 2024). We identify fundamental roadblocks and propose directions on how to make progress.