COMPASS: Contrastive Multimodal Pretraining for Autonomous Systems

  • Shuang Ma ,
  • Sai Vemprala ,
  • Wenshan Wang ,
  • Jayesh K. Gupta ,
  • Yale Song ,
  • Daniel McDuff ,
  • Ashish Kapoor

International Conference on Intelligent Robots and Systems |

Organized by IEEE

Learning representations that generalize across tasks and domains is challenging yet necessary for autonomous systems. Although task-driven approaches are appealing, designing models specific to each application can be difficult in the face of limited data, especially when dealing with highly variable multimodal input spaces arising from different tasks in different environments. We introduce the first general-purpose pretraining pipeline, COntrastive Multimodal Pretraining for AutonomouS Systems (COMPASS), to overcome the limitations of task-specific models and existing pretraining approaches. COMPASS constructs a multimodal graph by considering the essential information for autonomous systems and the properties of different modalities. Through this graph, multimodal signals are connected and mapped into two factorized spatiotemporal latent spaces: a “motion pattern space” and a “current state space.” By learning from multimodal correspondences in each latent space, COMPASS creates state representations that models necessary information such as temporal dynamics, geometry, and semantics. We pretrain COMPASS on a largescale multimodal simulation dataset TartanAir [1] and evaluate it on drone navigation, vehicle racing, and visual odometry tasks. The experiments indicate that COMPASS can tackle all three scenarios and can also generalize to unseen environments and real-world data.

论文与出版物下载

COMPASS

22 2 月, 2022

This repository contains the PyTorch implementation of the COMPASS model proposed in our paper: COMPASS: Contrastive Multimodal Pretraining for Autonomous Systems. COMPASS aims to build general purpose representations for autonomous systems from multimodal observations. Given multimodal signals of spatial and temporal modalities M_s and M_m, respectively. COMPASS learns two factorized latent spaces, i.e., a motion pattern space O_m and a current state space O_s, using multimodal correspondence as the self-supervisory signal.