Interactive and embodied tasks pose at least two fundamental challenges to existing Vision & Language (VL) models, including 1) grounding language in trajectories of actions and observations, and 2) referential disambiguation. To tackle these challenges, we propose an Embodied MultiModal Agent (EMMA): a unified encoder-decoder model that reasons over images and trajectories, and casts action prediction as multimodal text generation. By unifying all tasks as text generation, EMMA learns a language of actions which facilitates transfer across tasks. Different to previous modular approaches with independently trained components, we use a single multitask model where each task contributes to goal completion. EMMA performs on par with similar models on several VL benchmarks and sets a new state-of-the-art performance (36.81% success rate) on the Dialog-guided Task Completion (DTC), a benchmark to evaluate dialog-guided agents in the Alexa Arena
@article{arxiv.2311.04067,
title = {Multitask Multimodal Prompted Training for Interactive Embodied Task Completion},
author = {Georgios Pantazopoulos and Malvina Nikandrou and Amit Parekh and Bhathiya Hemanthage and Arash Eshghi and Ioannis Konstas and Verena Rieser and Oliver Lemon and Alessandro Suglia},
journal= {arXiv preprint arXiv:2311.04067},
year = {2023}
}