Tree of Thoughts enables language models to solve complex planning tasks by generating, evaluating, and searching over coherent intermediate thoughts in a tree, raising Game of 24 success from 4% to 74% with GPT-4.
hub Canonical reference
Inner Monologue: Embodied Reasoning through Planning with Language Models
Canonical reference. 81% of citing Pith papers cite this work as background.
75
Pith papers citing it
Background
81% of classified citations
abstract
Recent works have shown how the reasoning capabilities of Large Language Models (LLMs) can be applied to domains beyond natural language processing, such as planning and interaction for robots. These embodied problems require an agent to understand many semantic aspects of the world: the repertoire of skills available, how these skills influence the world, and how changes to the world map back to the language. LLMs planning in embodied environments need to consider not just what skills to do, but also how and when to do them - answers that change over time in response to the agent's own choices. In this work, we investigate to what extent LLMs used in such embodied contexts can reason over sources of feedback provided through natural language, without any additional training. We propose that by leveraging environment feedback, LLMs are able to form an inner monologue that allows them to more richly process and plan in robotic control scenarios. We investigate a variety of sources of feedback, such as success detection, scene description, and human interaction. We find that closed-loop language feedback significantly improves high-level instruction completion on three domains, including simulated and real table top rearrangement tasks and long-horizon mobile manipulation tasks in a kitchen environment in the real world.
hub tools
citation-role summary
background 23
baseline 1
dataset 1
extension 1
citation-polarity summary
claims ledger
- abstract Recent works have shown how the reasoning capabilities of Large Language Models (LLMs) can be applied to domains beyond natural language processing, such as planning and interaction for robots. These embodied problems require an agent to understand many semantic aspects of the world: the repertoire of skills available, how these skills influence the world, and how changes to the world map back to the language. LLMs planning in embodied environments need to consider not just what skills to do, but also how and when to do them - answers that change over time in response to the agent's own choice
co-cited works
representative citing papers
Generative agents with memory streams, reflection, and planning using LLMs exhibit believable individual and emergent social behaviors in a simulated town.
Language models generate robot policy code from natural language commands via few-shot prompting, enabling spatial-geometric reasoning, generalization, and precise control on real robots.
PromptPO shows LLMs can act as black-box policy optimizers for sequential RL when leveraging prior knowledge, matching baselines in exploration and robotics but underperforming in MuJoCo.
SDP constructs a task-induced state space from raw text by having agents commit to and certify natural-language predicates as states, enabling structured planning and analysis in unstructured language environments.
Action Agent pairs LLM-driven video generation with a flow-constrained diffusion transformer to produce velocity commands, raising video success to 86% and delivering 64.7% real-world navigation on a Unitree G1 humanoid.
LLM planners for robots often produce dangerous plans even when planning succeeds, with safety awareness staying flat as model scale improves planning ability.
ADAPT augments planners with affordance reasoning to raise task success in environments with unspecified and time-varying object affordances, and a LoRA-finetuned VLM backend beats GPT-4o on the new DynAfford benchmark.
ATLAS is the first silicon-validated simulation framework for 3D-DRAM LLM accelerators, achieving under 8.57% error and over 97% correlation with real hardware while supporting design exploration.
ST-BiBench reveals a coordination paradox in which MLLMs show strong high-level strategic reasoning yet fail at fine-grained 16-dimensional bimanual action synthesis and multi-stream fusion.
A survey that defines Compound AI Systems, proposes a multi-dimensional taxonomy based on component roles and orchestration strategies, reviews four foundational paradigms, and identifies key challenges for future research.
Orak is a foundational benchmark providing training data, interfaces, and evaluation tools for LLM agents across diverse video game genres.
SuSIE uses a finetuned InstructPix2Pix diffusion model to propose subgoal images that guide a low-level goal-conditioned policy, achieving SOTA zero-shot performance on CALVIN and real-world manipulation.
UniSim learns a universal real-world simulator from orchestrated diverse datasets, enabling zero-shot deployment of policies trained purely in simulation.
VoxPoser uses LLMs to compose 3D value maps via VLM interaction for model-based synthesis of robust robot trajectories on open-set language-specified manipulation tasks.
Voyager achieves superior lifelong learning in Minecraft by combining an automatic exploration curriculum, a library of executable skills, and iterative LLM prompting with environment feedback, yielding 3.3x more unique items and 15.3x faster milestone unlocks than prior methods while generalizing技能
LLM+P lets LLMs solve planning problems optimally by converting them to PDDL for classical planners and back to natural language.
SayCan combines an LLM's high-level semantic knowledge with robot skill value functions to select only feasible actions, enabling completion of abstract natural-language instructions on a real mobile manipulator.
InfoAtlas is a pretrained neural model for zero-shot mutual information estimation that matches state-of-the-art accuracy with 100x speedup and handles varying dimensions via a single model.
Any-ttach shows that rapid end-effector swapping combined with demonstration collection and task planning enables reliable multi-tool skills in long-horizon tasks such as sandwich making.
Introduces 9 synthetic annotation tasks and benchmarks for behavioral cloning, finding hierarchical skill learning, scaling benefits, effective multi-task pretraining, and shared internal representations of task phases and mistakes.
SPIKE dual-controller framework raises success rates 5-9 points and cuts tokens 55% in StarDojo agents by reusing strategic plans across stable segments and escalating only at detected events.
DeMiAn re-annotates robot and egocentric videos with VLM-generated dense labels across motion, scene, pose, and reasoning aspects, then uses a learned instructor to boost policy success by 5 points on RoboCasa over task-only baselines.
Ada-Diffuser is a causal diffusion model that jointly learns observed interaction structure and underlying latent dynamics from minimal observations for adaptive planning and policy learning.
citing papers explorer
-
When are LLMs Sufficient Policy Optimizers for Sequential RL Tasks?
PromptPO shows LLMs can act as black-box policy optimizers for sequential RL when leveraging prior knowledge, matching baselines in exploration and robotics but underperforming in MuJoCo.
-
State-Centric Decision Process
SDP constructs a task-induced state space from raw text by having agents commit to and certify natural-language predicates as states, enabling structured planning and analysis in unstructured language environments.
-
Action Agent: Agentic Video Generation Meets Flow-Constrained Diffusion
Action Agent pairs LLM-driven video generation with a flow-constrained diffusion transformer to produce velocity commands, raising video success to 86% and delivering 64.7% real-world navigation on a Unitree G1 humanoid.
-
Using large language models for embodied planning introduces systematic safety risks
LLM planners for robots often produce dangerous plans even when planning succeeds, with safety awareness staying flat as model scale improves planning ability.
-
ADAPT: Benchmarking Commonsense Planning under Unspecified Affordance Constraints
ADAPT augments planners with affordance reasoning to raise task success in environments with unspecified and time-varying object affordances, and a LoRA-finetuned VLM backend beats GPT-4o on the new DynAfford benchmark.
-
A Full-Stack Performance Evaluation Infrastructure for 3D-DRAM-based LLM Accelerators
ATLAS is the first silicon-validated simulation framework for 3D-DRAM LLM accelerators, achieving under 8.57% error and over 97% correlation with real hardware while supporting design exploration.
-
ST-BiBench: Benchmarking Multi-Stream Multimodal Coordination in Bimanual Embodied Tasks for MLLMs
ST-BiBench reveals a coordination paradox in which MLLMs show strong high-level strategic reasoning yet fail at fine-grained 16-dimensional bimanual action synthesis and multi-stream fusion.
-
InfoAtlas: A Foundation Model for Zero-Shot Statistical Dependence Estimate
InfoAtlas is a pretrained neural model for zero-shot mutual information estimation that matches state-of-the-art accuracy with 100x speedup and handles varying dimensions via a single model.
-
Any-ttach: Quick End-effector Swapping Enables Manipulation Dexterity with Simplicity
Any-ttach shows that rapid end-effector swapping combined with demonstration collection and task planning enables reliable multi-tool skills in long-horizon tasks such as sandwich making.
-
A Systematic Study of Behavioral Cloning for Scientific Data Annotation
Introduces 9 synthetic annotation tasks and benchmarks for behavioral cloning, finding hierarchical skill learning, scaling benefits, effective multi-task pretraining, and shared internal representations of task phases and mistakes.
-
SPIKE: An Adaptive Dual Controller Framework for Cost-Efficient Long-Horizon Game Agents
SPIKE dual-controller framework raises success rates 5-9 points and cuts tokens 55% in StarDojo agents by reusing strategic plans across stable segments and escalating only at detected events.
-
How to Instruct Your Robot: Dense Language Annotations Power Robot Policy Learning
DeMiAn re-annotates robot and egocentric videos with VLM-generated dense labels across motion, scene, pose, and reasoning aspects, then uses a learned instructor to boost policy success by 5 points on RoboCasa over task-only baselines.
-
Ada-Diffuser: Latent-Aware Adaptive Diffusion for Decision-Making
Ada-Diffuser is a causal diffusion model that jointly learns observed interaction structure and underlying latent dynamics from minimal observations for adaptive planning and policy learning.
-
Towards Long-horizon Embodied Agents with Tool-Aligned Vision-Language-Action Models
VLAs-as-Tools pairs a VLM planner with specialized VLA executors via a new interface and Tool-Aligned Post-Training to raise long-horizon robot success rates on LIBERO-Long and RoboTwin benchmarks.
-
From Reaction to Anticipation: Proactive Failure Recovery through Agentic Task Graph for Robotic Manipulation
AgentChord models manipulation tasks as directed graphs enriched with anticipatory recovery branches, using specialized agents to enable immediate, low-latency failure responses and improve success on long-horizon bimanual tasks.
-
From Controlled to the Wild: Evaluation of Pentesting Agents for the Real-World
A practical evaluation protocol for AI pentesting agents that uses validated vulnerability discovery, LLM semantic matching, and bipartite scoring to assess performance in realistic, complex targets.
-
RePO-VLA: Recovery-Driven Policy Optimization for Vision-Language-Action Models
RePO-VLA raises average adversarial success rates in VLA manipulation from 20% to 75% by using recovery-aware initialization, a progress-aware semantic value function, and value-conditioned refinement on success and corrective trajectories.
-
OPT-BENCH: Evaluating the Iterative Self-Optimization of LLM Agents in Large-Scale Search Spaces
OPT-BENCH and OPT-Agent evaluate LLM self-optimization in large search spaces, showing stronger models improve via feedback but stay constrained by base capacity and below human performance.
-
Milestone-Guided Policy Learning for Long-Horizon Language Agents
BEACON uses milestone partitioning, temporal reward shaping, and dual-scale advantage estimation to nearly double success rates on long-horizon ALFWorld tasks while raising effective sample use from 23.7% to 82%.
-
LoopTrap: Termination Poisoning Attacks on LLM Agents
LoopTrap is an automated red-teaming framework that crafts termination-poisoning prompts to amplify LLM agent steps by 3.57x on average (up to 25x) across 8 agents.
-
How Far Are VLMs from Privacy Awareness in the Physical World? An Empirical Study
Vision-language models exhibit perceptual fragility and fail to consistently respect privacy constraints when operating in simulated physical environments, with performance declining in cluttered scenes and under conflicting commands.
-
Stable Agentic Control: Tool-Mediated LLM Architecture for Autonomous Cyber Defense
Tool-mediated LLM agents with deterministic tools and a machine-checked Lyapunov certificate achieve stable control in cyber defense, reducing attacker game value by 59% on real attack graphs.
-
FlexSQL: Flexible Exploration and Execution Make Better Text-to-SQL Agents
FlexSQL reaches 65.4% on Spider2-Snow by allowing agents to flexibly explore schemas, generate diverse plans, choose SQL or Python execution, and apply two-tiered repair.
-
Efficient Rationale-based Retrieval: On-policy Distillation from Generative Rerankers based on JEPA
Rabtriever distills a generative reranker into an efficient bi-encoder using on-policy JEPA to achieve near-reranker accuracy with linear complexity on rationale-based retrieval.
-
Long-Horizon Manipulation via Trace-Conditioned VLA Planning
LoHo-Manip enables robust long-horizon robot manipulation by using a receding-horizon VLM manager to output progress-aware subtask sequences and 2D visual traces that condition a VLA executor for automatic replanning.
-
Navigating the Clutter: Waypoint-Based Bi-Level Planning for Multi-Robot Systems
Waypoint-based bi-level planning with curriculum RLVR improves multi-robot task success rates in dense-obstacle benchmarks over motion-agnostic and VLA baselines.
-
HELM: Harness-Enhanced Long-horizon Memory for Vision-Language-Action Manipulation
HELM raises long-horizon VLA success from 58.4% to 81.5% on LIBERO-LONG by combining episodic memory retrieval, learned failure prediction, and replanning, outperforming context extension or adaptation alone.
-
From Perception to Planning: Evolving Ego-Centric Task-Oriented Spatiotemporal Reasoning via Curriculum Learning
EgoTSR applies a three-stage curriculum on a 46-million-sample dataset to build egocentric spatiotemporal reasoning, reaching 92.4% accuracy on long-horizon tasks and reducing chronological biases.
-
An Edge-Host-Cloud Architecture for Robot-Agnostic, Caregiver-in-the-Loop Personalized Cognitive Exercise: Multi-Site Deployment in Dementia Care
Speaking Memories is a robot-agnostic edge-host-cloud architecture for caregiver-in-the-loop personalized cognitive exercise in dementia care, achieving sub-6-second latency and positive stakeholder feedback in multi-site deployments.
-
BrainMem: Brain-Inspired Evolving Memory for Embodied Agent Task Planning
BrainMem equips LLM-based embodied planners with working, episodic, and semantic memory that evolves interaction histories into retrievable knowledge graphs and guidelines, raising success rates on long-horizon 3D benchmarks.
-
Tracking Capabilities for Safer Agents
AI agents can generate code in a capability-safe Scala dialect that statically prevents information leakage and malicious side effects while preserving task performance.
-
SoK: Agentic Skills -- Beyond Tool Use in LLM Agents
The paper systematizes agentic skills beyond tool use, providing design pattern and representation-scope taxonomies plus security analysis of malicious skill infiltration in agent marketplaces.
-
PALM: Progress-Aware Policy Learning via Affordance Reasoning for Long-Horizon Robotic Manipulation
PALM improves long-horizon robotic manipulation success by distilling affordance representations for object interaction and predicting within-subtask progress in a VLA model.
-
SWEET: Sparse World Modeling with Image Editing for Embodied Task Execution
SWEET is a one-shot sparse visual planning framework that progressively generates manipulation keyframes via image editing conditioned on language and spatial guidance, then converts them to actions with a diffusion predictor, showing better fidelity and lower cost than video models on DROID and Rob
-
TaskGround: Structured Executable Task Inference for Full-Scene Household Reasoning
TaskGround introduces a Ground-Infer-Execute framework for full-scene household reasoning that improves success rates on the FullHome benchmark and enables compact models to match larger ones at up to 18x lower token cost.
-
Large Language Models for Sequential Decision-Making: Improving In-Context Learning via Supervised Fine-Tuning
Supervised fine-tuning of pretrained LLMs on offline trajectories yields better few-shot sequential decision-making than in-context-only baselines, with a theoretical suboptimality bound derived for linear MDPs by interpreting attention as Q-function estimation.
-
Evaluating Generative Models as Interactive Emergent Representations of Human-Like Collaborative Behavior
Embodied LLM agents exhibit emergent collaborative behaviors indicating mental models of partners in a color-matching game, detected via LLM judges and supported by positive user feedback.
-
Pre-Execution Safety Gate & Task Safety Contracts for LLM-Controlled Robot Systems
SafeGate adds a deterministic pre-execution gate and runtime contracts with Z3 SMT solving to block unsafe LLM commands for robots while passing safe ones.
-
From Actions to Understanding: Conformal Interpretability of Temporal Concepts in LLM Agents
A conformal interpretability method labels LLM agent states step-by-step and extracts linearly separable temporal concept directions aligned with task success on ScienceWorld and AlfWorld.
-
LightZeroNav: Zero-Shot Vision Language Navigation in Continuous Environments Based on Lightweight VLMs
LightZeroNav decomposes zero-shot VLN-CE into modules that reduce input redundancy, improve progress tracking from noisy memory, and separate action execution from stage transitions, allowing an 8B VLM to match GPT-4o performance.
-
A Hierarchical Error-Corrective Graph Framework for Autonomous Agents with LLM-Based Action Generation
HECG combines multi-dimensional metrics for strategy choice, ten-type error classification with recoverability details, and causal-context graphs to improve LLM agent reliability in complex tasks.
-
Embodied Task Planning via Graph-Informed Action Generation with Large Language Models
GiG uses a Graph-in-Graph architecture with GNN-encoded states, experience memory retrieval, and bounded symbolic lookahead to improve LLM planning on embodied benchmarks with gains up to 37%.
-
ABot-Claw: A Foundation for Persistent, Cooperative, and Self-Evolving Robotic Agents
ABot-Claw is an embodied software layer that adds unified robot scheduling, cross-embodiment visual memory, and critic-driven replanning on top of OpenClaw to support persistent multi-robot execution from natural-language goals.
-
Agentic Reasoning for Large Language Models
The survey structures agentic reasoning for LLMs into foundational, self-evolving, and collective multi-agent layers while distinguishing in-context orchestration from post-training optimization and reviewing applications across domains.
- 2.5-D Decomposition for LLM-Based Spatial Construction
- Governed Capability Evolution: Lifecycle-Time Compatibility Checking and Rollback for AI-Component-Based Systems, with Embodied Agents as Case Study
- Harnessing Embodied Agents: Runtime Governance for Policy-Constrained Execution