LLM guided search improves embodied task planning in complex environments
TESS: LLM-guided Translation, Exploration, Simulation, and Search for Embodied Task Planning
Artificial Intelligence
Summary
Planning tasks that involve physical actions can be tricky because they need to be broken down into small steps, and it’s hard to make sure these steps are both correct and doable. The authors found that while traditional search methods are perfect in theory, they need complete information, which isn't usually available. To tackle this, they created a method that uses large language models to help with understanding instructions, exploring the environment, simulating actions, and narrowing down choices. Their experiments show that this new approach works better on popular simulation platforms.
What this means in practice
- •For robotic system developers: Create robots that translate human instructions into clear goals and efficiently plan their actions in partially known environments using LLM guided search.
- •For virtual assistant builders: Design virtual agents that better interpret user commands and simulate possible outcomes to complete complex tasks in virtual home environments.
Authors
Peiran Xu, Yixiang Yang, Jinghan Li, Fei Li, Cao Sheng, JiaZhong Yu, Yadong Mu
Abstract
In this work we focus on the problem of embodied task planning, which requires an agent to accomplish a complex task by separating it into atomic actions. Recent studies have developed LLM-based methods to address this problem. However, the executability and correctness of the generated plans remain suboptimal in complex environments and tasks. On the other hand, classical search algorithms have the potential to solve the planning problems perfectly, yet they require fully observable environments, known state transitions and goal state, which are often unattainable in general task domains. Therefore, we propose to implement an LLM-guided search algorithm that makes the search process possible and efficient by incorporating LLMs' commonsense knowledge. Our method leverages LLMs to (1) translate natural language instructions into formatted goal conditions, (2) facilitate efficient exploration in partially observable environments, (3) implement a rule-based world simulator, and (4) prune the space of candidate actions to improve search efficiency. Experimental results on VirtualHome and ALFWorld demonstrate its superior performance.