Embodied AI systems are increasingly deployed in open-world environments, yet ensuring their reliability remains a fundamental challenge. Drawing on discussions from the AAAI’26 Bridge Program on “Making Embodied AI Reliable with Testing and Formal Verification”, this article argues that reliability in embodied AI is inherently a lifecycle assurance problem arising from uncertainty, human interaction, and emergent behaviors across tightly coupled system components. We identify three complementary directions toward reliable embodied AI: (i) trustworthy scenario-based testing grounded in validated specifications and coverage-driven evaluation, (ii) compositional verification enabled through structured symbolic abstractions, and (iii) runtime assurance mechanisms capable of adapting under uncertainty and distribution shifts. Rather than treating these approaches independently, we advocate for integrated assurance workflows that connect testing, verification, and runtime adaptation through shared neuro-symbolic representations and closed-loop feedback across the system lifecycle. Such integration provides a foundation for trustworthy embodied AI systems operating safely in complex real-world environments.

@inproceedings{ZhengArxiv26,
  title     = {Making Embodied AI Reliable: A Community Agenda from Testing to Formal Verification},
  author    = {Zheng, Xi and Weerakoon, Dulanga and Huo, Yintong and Yeo, Teresa and Van den Broeck, Guy and Ganesh, Vijay and Neider, Daniel and Srivastava, Biplav and Ruchkin, Ivan and Misra, Archan and Pasareanu, Corina},
  booktitle = {Arxiv},
  url       = "https://starai.cs.ucla.edu/papers/ZhengArxiv26.pdf",
  eprint    = {2606.03593},
  archivePrefix = {arXiv},
  primaryClass = {cs.SE},
  month     = 6,
  year      = {2026},
  keywords  = {techreport}
}