aiqbench analysis
This paper introduces COMET, a reinforcement learning algorithm that integrates causal object-centric modeling into Monte Carlo Tree Search for more efficient planning in complex environments. By binding actions to specific objects through novel fusion mechanisms and using object-causal attention, the method improves early-stage training performance across diverse robotic benchmarks compared to monolithic baselines.
Causal Object-Centric Models for Planning with Monte Carlo Tree Search
arXiv:2606.14418v1 Announce Type: cross Abstract: We introduce COMET (Causal Object-centric Model for Efficient Tree search), a model-based reinforcement learning algorithm that performs Monte Carlo Tree Search in a slot-structured latent space. COMET pairs a frozen unsupervised object-centric encoder with a transformer-based world model, in which actions are bound to objects through a novel action-slot fusion mechanism that is used in slot transition prediction. Policy and value heads use object-causal attention, modulating token interactions by learned per-slot relevance scores so that decision-making concentrates on task-relevant entities. COMET adds an explicit object-level inductive bias to MuZero-style latent planning. Across eight visually and dynamically diverse tasks from the Object-Centric Visual RL benchmark, ManiSkill, Robosuite, and VizDoom, COMET achieves a higher mean normalized score during the early stages of training compared to object-centric and monolithic baselines.
Abstract:We introduce COMET (Causal Object-centric Model for Efficient Tree search), a model-based reinforcement learning algorithm that performs Monte Carlo Tree Search in a slot-structured latent space. COMET pairs a frozen unsupervised object-centric encoder with a transformer-based world model, in which actions are bound to objects through a novel action-slot fusion mechanism that is used in slot transition prediction. Policy and value heads use object-causal attention, modulating token interactions by learned per-slot relevance scores so that decision-making concentrates on task-relevant entities. COMET adds an explicit object-level inductive bias to MuZero-style latent planning. Across eight visually and dynamically diverse tasks from the Object-Centric Visual RL benchmark, ManiSkill, Robosuite, and VizDoom, COMET achieves a higher mean normalized score during the early stages of training compared to object-centric and monolithic baselines.
Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
arXiv:2606.14418 [cs.AI]
(or arXiv:2606.14418v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2606.14418 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 226 words
Related coverage
Adaptive Parallel Reasoning: The Next Paradigm in Efficient Inference Scaling
BAIR Blog · May 8, 2026
A Mike's-Eye View of ARC's Research
AI Alignment Forum · Jun 9, 2026
Generative Frontier Planning for Adaptive Peer-Referral Recruitment under Covariate-Dependent Arrivals
arXiv · Jun 9, 2026
DiffuSearch: How Hybrid Trajectory Planning Benefits from Aligned Objectives in Diffusion and Action Space
arXiv · Sep 3, 2026