Subbarao Kambhampati
Professor, School of Computing & AI, Arizona State University
WebsiteTentative talk title
About the workshop
NEmo focuses on embodied neuro-symbolic AI: the integration of learned perception, reinforcement learning, large language models, planning, and structured symbolic knowledge in agents that act in physical or richly simulated environments.
Robotic and interactive agents must perceive changing worlds, reason over goals and constraints, execute long action sequences, and recover when assumptions fail. In this setting, hallucinated preconditions, brittle domain models, and ungrounded affordance knowledge become safety and reliability problems.
The workshop is designed for researchers in machine learning, knowledge representation, planning, robotics, human-robot interaction, semantic web, and trustworthy AI who are working on adjacent parts of the same problem.
Venue
NeurIPS 2026
Date
December 12, 2026
Location
Sydney, Australia
Core Challenges
The programme is organised around three technical challenges.
Long-Horizon Planning
Interfaces for learning, revising, and verifying symbolic action models while learned components handle perception and low-level control.
LLMs and Provenance
Reliable use of LLMs as domain-model elicitors, policy priors, symbolic constraint generators, and natural-language interfaces.
Shared Semantics
Reusable knowledge layers connecting environments, object affordances, robot capabilities, plans, and execution traces.
Topics
- Neuro-symbolic AI for embodied agents
- LLMs for planning in atypical domains
- Knowledge representation for planning
- Ontological reasoning for embodied applications
- Tacit planning knowledge in frontier models
- Symbolic model elicitation for embodied reasoning
- Perception and sensor grounding
- Rule-based validation of symbolic domains
- Neuro-symbolic reinforcement learning
- Long-horizon planning and plan repair
- Self-evolving agents
- Ethical aspects of embodied AI systems
Workshop Format
A one-day, in-person workshop of roughly 8.5 hours.
Sessions
The format combines keynotes, paper presentations, posters, demos, and structured breakout discussions. Organizers will introduce and moderate; keynotes will come from external speakers.
The goal is to leave the day with a short public report covering open problems, shared assumptions, and candidate benchmarks or semantic layers.
Demos are expected to be laptop-, video-, or browser-based unless additional equipment is secured.
Session I: Opening, Keynote I, and Papers
Opening, goals, and audience poll
Keynote I
Paper presentations I
Session II: Posters, Demos, Keynote II, and Papers
Coffee, posters, and demos I
Keynote II
Paper presentations II
Session III: Lunch, Keynote III, Papers, and Posters
Lunch
Keynote III
Paper presentation III
Coffee, posters, and demos II
Session IV: Papers, Breakouts, and Closing
Paper presentations IV
Structured breakout discussions
Closing and best contribution award
Invited Speakers
Meet the invited speakers joining NEmo 2026.
Accepted Papers
8 oral presentations and 30 posters.
Camera-ready PDFs will be available after the camera-ready deadline (31 October 2026).
Oral Presentations
-
Discrete Actions for Naturalistic Behavior in Embodied Biomechanical Animal Models
-
House Rules: A Benchmark for Agents under Changing Requirements
-
Iterative Policy Refinement through Semantic Rollout Analysis
-
LLM-Symbolic PDDL Model Repair
-
Meta-Ctrl: Guaranteed Plan Generation by Decoupling Syntactic and Semantic Constraints
-
MindTopo: Can Foundation Models Reason in Topological Space?
-
Safe Reinforcement Learning via Probabilistic Option Shielding
-
Semantic Parsing and Grounding of Assembly Instructions With Ontologies
Posters
-
Answering Without Arriving: An Evidence Audit for Embodied Question Answering
-
ARIS: Agentic and Relationship Intelligence System for Social Robots
-
Causal Calibration of Symbolic State in Embodied Language Agents
-
DISEIL: Demonstration Distillation for Sample-Efficient Imitation Learning
-
Embodiment Breaks the Tie: Identifying Whether the Model or the Grounder Failed in Neuro-Symbolic Planning
-
Evaluating LLM-Generated World Models for Neuro-Symbolic Reasoning
-
Four Planners, One Building: Shared-Task Benchmarking of Neural and Neurosymbolic World Models, and Three Ways It Misleads
-
From Embodied Contingency to Arbitrary Symbols? A Diagnose-Repair-Retest Benchmark
-
Human-Grounded Representations for Neuro-Symbolic Monitoring of Multi-Agent Coordination
-
Learning Symbolic Task Grounding for Temporally Structured Reinforcement Learning
-
LeVLWM: Text-Goal Control with a Single Vision-Language World Model
-
Managing Action Preconditions in Neuro-Symbolic RL: Three Placement Strategies for Embodied Agents
-
Minimal Particle Set Tracking with Neurosymbolic State Estimation for Robust Planning and Execution
-
ModelForge: Test-Guided Generation and Repair of Planning Domains from Natural Language
-
Neuro-Symbolic Validation for Multi-Robot Task Updates from Human Language
-
Observable Symbolic Validation and Capacity-Aware Recovery for Safer Vision-Language Robotic Sorting
-
OCC4M: Grounded Object-Centric Memory for Neuro-Symbolic Manipulation
-
Planning Guided Option Critic
-
Policy-Induced Blind Spots in Symbolic Action-Model Repair
-
PRAXIS over Theoria: Self-Exposure in Place of Scale for Efficient Manipulation
-
Prove Before You Move: Retrieval-Augmented Deduction for Embodied Agents
-
Recursive World-Model Self-Repair in PDDL Domains: Verifier Coverage Is the Binding Constraint, and Under-Verified Loops Stall Rather Than Diverge
-
Relevance-Filtered Symbolic Context Remediates the Task-Planning Bottleneck in Embodied LLM Agents
-
Roz: A Taste Test for World Model Latents
-
Symbolic Command Grounding over Frozen Neural Perception: When Does a Rule Layer Help Referring Segmentation?
-
Synthesis of Shields for Safe Reinforcement Learning in Both Discrete and Continuous State and Action Spaces with Linear Dynamics
-
Track-and-Complete: Learning Humanoids Skills from a Single Failed Human Video
-
Verifiable Step-Wise Planning: Cost-to-Go Reward Design for Language Model Planners
-
What Does an Assembly Monitor Need to Remember?
-
Which Predictions Are Worth Updating? Budgeted Predictive Knowledge for Long-Horizon Agents
Organizing Committee
The team spans robotics, planning, reinforcement learning, neuro-symbolic AI, semantic web, and knowledge graphs.
Call for Papers
Contributed work is non-archival and managed through OpenReview.
Long Papers
- Up to 8 pages, excluding references and appendices
- Original work on embodied neuro-symbolic AI
- Three reviews per submission
Short Papers
- Up to 4 pages, excluding references and appendices
- Position papers, early results, benchmarks, demos
- Eligible for lightning talks and posters
Submission and Review
- Submit papers through OpenReview.
- Use the official NeurIPS 2026 LaTeX template. For workshop submissions, use the single-blind workshop option,
sglblindworkshop. - Submissions will be reviewed single-blind by the programme committee: reviewers will see author identities, while reviewer identities will remain confidential.
- NEmo is a non-proceedings workshop. Accepted papers will be published on the workshop website.
- Dual-submission policy: The workshop will adopt a non-archival policy, welcoming ongoing and unpublished work, as well as papers under review or recently accepted at other venues. We discourage the submission of work previously published at major venues (e.g., NeurIPS, ICLR, ICML).
Important Dates
Workshop timeline.
- Submission deadline:
September 10, 2026September 14, 2026 (AoE) - Notifications sent: September 29, 2026
- Camera-ready paper deadline: October 31, 2026
- Workshop date: December 12, 2026 (Sydney)
Contact
Venue
NeurIPS 2026, Sydney, AustraliaContact Person
Emanuele Musumeci