Gibson Env: Real-World Perception for Embodied Agents

TL;DR

Gibson Env is a virtual environment that models real-world spaces to train perception models for embodied agents, enabling transfer to real-world applications without additional domain adaptation.

Contribution

The paper introduces Gibson Env, a virtual platform based on real spaces, facilitating perception learning for agents with direct transfer capabilities to real-world scenarios.

Findings

Gibson includes over 1400 real-world floor spaces from 572 buildings.

The environment supports training perception models that transfer directly to real-world robots.

Sample perceptual tasks demonstrate effective real-world deployment without domain adaptation.

Abstract

Developing visual perception models for active agents and sensorimotor control are cumbersome to be done in the physical world, as existing algorithms are too slow to efficiently learn in real-time and robots are fragile and costly. This has given rise to learning-in-simulation which consequently casts a question on whether the results transfer to real-world. In this paper, we are concerned with the problem of developing real-world perception for active agents, propose Gibson Virtual Environment for this purpose, and showcase sample perceptual tasks learned therein. Gibson is based on virtualizing real spaces, rather than using artificially designed ones, and currently includes over 1400 floor spaces from 572 full buildings. The main characteristics of Gibson are: I. being from the real-world and reflecting its semantic complexity, II. having an internal synthesis mechanism, "Goggles", enabling deploying the trained models in real-world without needing further domain adaptation, III. embodiment of agents and making them subject to constraints of physics and space.