Active Neural Mapping

TL;DR

This paper introduces Active Neural Mapping, a novel system that actively explores environments using neural scene representations, leveraging neural variability to guide exploration and improve scene reconstruction efficiency.

Contribution

It is the first to use coordinate-based implicit neural representations for online active scene mapping and employs neural variability to measure map uncertainty.

Findings

Effective in reducing map uncertainty during exploration

Demonstrates superior performance in Gibson and Matterport3D environments

Enables online scene reconstruction with neural representations

Abstract

We address the problem of active mapping with a continually-learned neural scene representation, namely Active Neural Mapping. The key lies in actively finding the target space to be explored with efficient agent movement, thus minimizing the map uncertainty on-the-fly within a previously unseen environment. In this paper, we examine the weight space of the continually-learned neural field, and show empirically that the neural variability, the prediction robustness against random weight perturbation, can be directly utilized to measure the instant uncertainty of the neural map. Together with the continuous geometric information inherited in the neural map, the agent can be guided to find a traversable path to gradually gain knowledge of the environment. We present for the first time an active mapping system with a coordinate-based implicit neural representation for online scene reconstruction. Experiments in the visually-realistic Gibson and Matterport3D environment demonstrate the efficacy of the proposed method.