PaLM-E: An Embodied Multimodal Language Model

TL;DR

PaLM-E is a large embodied multimodal language model that integrates visual, sensor, and textual data to perform diverse real-world tasks like robotics, visual question answering, and captioning, demonstrating positive transfer and state-of-the-art performance.

Contribution

This work introduces PaLM-E, a unified large-scale embodied multimodal language model that combines sensor data with language understanding for complex real-world tasks.

Findings

PaLM-E achieves state-of-the-art results on OK-VQA.

The model benefits from joint training across multiple modalities.

PaLM-E retains general language capabilities at scale.

Abstract

Large language models excel at a wide range of complex tasks. However, enabling general inference in the real world, e.g., for robotics problems, raises the challenge of grounding. We propose embodied language models to directly incorporate real-world continuous sensor modalities into language models and thereby establish the link between words and percepts. Input to our embodied language model are multi-modal sentences that interleave visual, continuous state estimation, and textual input encodings. We train these encodings end-to-end, in conjunction with a pre-trained large language model, for multiple embodied tasks including sequential robotic manipulation planning, visual question answering, and captioning. Our evaluations show that PaLM-E, a single large embodied multimodal model, can address a variety of embodied reasoning tasks, from a variety of observation modalities, on multiple embodiments, and further, exhibits positive transfer: the model benefits from diverse joint training across internet-scale language, vision, and visual-language domains. Our largest model, PaLM-E-562B with 562B parameters, in addition to being trained on robotics tasks, is a visual-language generalist with state-of-the-art performance on OK-VQA, and retains generalist language capabilities with increasing scale.