CFP - Architectures for Embodied Agents (AAAI 2026 Summer Symposium)

[Dongkyu Choi]

CFP - AAAI 2026 Summer Symposium Series

Architectures for Embodied Agents: A Synergy of Classic and Foundation Model Paradigms

Dear Colleagues,

We are pleased to announce the Call for Papers for the symposium on “Architectures for Embodied Agents: A Synergy of Classic and Foundation Model Paradigms” to be held June 22 - June 24, 2026 at Dongguk University in Seoul, South Korea.

 

About the Symposium

The development of reliable, verifiable, and safe embodied agents is a major challenge for modern AI. The field's evolution spans from early structured systems such as cognitive architectures with formal guarantees and robotic architectures with precise control, to today's foundation models, which bring flexible perception and unprecedented generality.

 

However, each approach has critical limitations: cognitive and robotic architectures often lack the flexible, open-ended knowledge required to operate effectively in complex, novel environments, while foundation model-based systems are often unpredictable due to hallucinations and lack verifiable safety checks, owing to their black-box nature, which hampers their application in mission-critical settings.

 

This symposium aims to address these critical challenges by exploring the essential role of control architectures. By gathering experts from diverse fields, we can define principles for robust, long-term agent control. The focus will be on architectures that provide reliable frameworks for imposing logical constraints, enforcing safety checks, handling reasoning over physical state, and translating system outputs into safe, executable commands and actions. Our goal is to integrate the reliability of classic, structured approaches with the generative power of modern foundation models, creating a new generation of trustworthy, embodied intelligence.

 

Topics

We invite contributions that explore how these different approaches can work together to create reliable embodied agents. Areas of interest include:

Core Methodologies: Hybrid architectures that combine reasoning and planning with broad-coverage data-driven expertise; verifiable methods for applying formal methods to stochastic outputs; and use of explicit models or knowledge to guide agent behaviour.

Applications: Reliable long-horizon planning in robotics and autonomous systems; systems for safety-critical domains like manufacturing and driving; and human-robot teams that collaborate safely.

Interdisciplinary Connections: Inspirations from cognitive architectures; frameworks for ethical enforcement; and system design for explainable embodied agents.

 

Format

The symposium will be a two-and-a-half-day interactive event. The format will include two to three invited talks from leading experts, contributed talks based on accepted abstracts, and poster sessions for informal discussion and networking. The target audience includes researchers, practitioners, and graduate students from Robotic Systems, Cognitive Systems, Safe AI, Human-Robot Interaction, and Systems Engineering.

 

Submission Requirements

Authors should submit two-page abstracts that will be evaluated based on relevance and potential to spark discussion. Authors can choose to have their work published in the AAAI Summer Symposia Proceedings, as long as the final version is submitted by April 15, 2026.

Submission site: https://easychair.org/conferences?conf=suss26

Submission deadline: May 1, 2026

(April 1 for inclusion in AAAI Summer Symposia Proceedings)

 

More information can be found at https://aea-aaai2026.com