[NeurIPS 2023] We use large language models as commonsense world model and heuristic policy within Monte-Carlo Tree Search, enabling better-reasoned decision-making for daily task planning problems.

[NeurIPS 2024] Official implementation for paper "Can Graph Learning Improve Planning in LLM-based Agents?"

Integrating learning and task planning for robots with Keras, including simulation, real robot, and multiple dataset support.

Human Demo Videos to Robot Action Plans

使用langchain进行任务规划，构建子任务的会话场景资源，通过MCTS任务执行器，来让每个子任务通过在上下文中资源，通过自身反思探索来获取自身对问题的最优答案；这种方式依赖模型的对齐偏好，我们在每种偏好上设计了一个工程框架，来完成自我对不同答案的奖励进行采样策略

Task management system

Planning as In-Painting: A Diffusion-Based Embodied Task Planning Framework for Environments under Uncertainty

Official implementation for the LABOR (LAnguage-model-based Bimanual ORchestration) Agent.

[CoRL 2022] BusyBot: Learning to Interact, Reason, and Plan in a BusyBoard Environment

Code for the ICRA 2024 Cook2LTL paper on translating free-form cooking recipes to Linear Temporal Logic (LTL) formulae for robot task planning.

The hierarchical planner CHIMP (Conflict-driven Hierarchical Meta-CSP Planner).

TME: Structured memory engine for LLM agents to plan, rollback, and reason across multi-step tasks. DAG upgrade in progress.

Kanban Pro: Streamline Your Workflow 🥰 Please Give Me A Star ⭐️

Code for the paper "Differentiable Task Graph Learning: Procedural Activity Representation and Online Mistake Detection from Egocentric Videos" [NeurIPS (spotlight), 2024]

Code for Transformers are Adaptable Task Planners, CoRL 2022

Task Allocation Strategy to distribute goals (in a decentralised manner) in a heterogeneous fleet considering, goals makespan, goals position, robot capabilities.

[ICML 2025] Closed-Loop Long-Horizon Robotic Planning via Equilibrium Sequence Modeling

Official implementation of LLM+MAP: Bimanual Robot Task Planning using Large Language Models (LLMs) and Planning Domain Definition Language (PDDL). Codes and files are coming soon.

Hybrid ai agent framework

Planning domains motivated by real-world problems, particularly robotics applications, which contain incomplete knowledge, temporal and numeric requirements.