Wonderful Team: Zero-Shot Physical Task Planning with Visual LLMs

TL;DR

Wonderful Team leverages Vision Large Language Models for zero-shot high-level robotic planning directly from environment images, outperforming previous methods by integrating perception, control, and planning.

Contribution

The paper introduces a novel multi-agent VLLM framework for zero-shot robotic planning that eliminates the need for separate vision systems, enabling more integrated and effective high-level task execution.

Findings

40% success rate improvement on VimaBench

30% improvement over Trajectory Generators on drawing and wiping tasks

70% improvement on semantic reasoning tasks with linguistic constraints

Abstract

We introduce Wonderful Team, a multi-agent Vision Large Language Model (VLLM) framework for executing high-level robotic planning in a zero-shot regime. In our context, zero-shot high-level planning means that for a novel environment, we provide a VLLM with an image of the robot's surroundings and a task description, and the VLLM outputs the sequence of actions necessary for the robot to complete the task. Unlike previous methods for high-level visual planning for robotic manipulation, our method uses VLLMs for the entire planning process, enabling a more tightly integrated loop between perception, control, and planning. As a result, Wonderful Team's performance on real-world semantic and physical planning tasks often exceeds methods that rely on separate vision systems. For example, we see an average 40% success rate improvement on VimaBench over prior methods such as NLaP, an average 30% improvement over Trajectory Generators on tasks from the Trajectory Generator paper, including drawing and wiping a plate, and an average 70% improvement over Trajectory Generators on a new set of semantic reasoning tasks including environment rearrangement with implicit linguistic constraints. We hope these results highlight the rapid improvements of VLLMs in the past year, and motivate the community to consider VLLMs as an option for some high-level robotic planning problems in the future.