Are Transformers Truly Foundational for Robotics?

TL;DR

This paper critically examines the hype around GPTs for robotics, highlighting their high resource costs and contrasting them with biological systems like insect brains that achieve robust autonomy efficiently.

Contribution

It challenges the assumption that GPTs are inherently suitable for robotics and suggests biological insights can improve robotic AI.

Findings

GPTs require enormous compute and training time

Biological systems achieve autonomy with minimal resources

Lessons from insect brains can inform better robotic AI

Abstract

Generative Pre-Trained Transformers (GPTs) are hyped to revolutionize robotics. Here we question their utility. GPTs for autonomous robotics demand enormous and costly compute, excessive training times and (often) offboard wireless control. We contrast GPT state of the art with how tiny insect brains have achieved robust autonomy with none of these constraints. We highlight lessons that can be learned from biology to enhance the utility of GPTs in robotics.