Posted on :: Tags: , , , , , , , , ,

Table of Contents

Introduction

In this blog, we try to train a text-conditioned robot control policy. It extends a naive adversarial imitation policy by supporting user text instructions to generate motions with different styles. Specifically, we integrate a BERT language model into an adversarial imitation RL pipeline.

Results

Text
CMD: Atk Kick
CMD: Atk Jump
CMD: Atk Shield Charge
CMD: Atk Shield Swipe
CMD: Dodge Backward
CMD: Atk Shield Charge
CMD: Atk Slash Up
CMD: Atk Slash Down
CMD: Atk Slash Left
CMD: Atk Slash Right
CMD: Atk Combo

Comments