Multi-Object Navigation with dynamically learned neural implicit representations - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2023

Multi-Object Navigation with dynamically learned neural implicit representations

Abstract

Understanding and mapping a new environment are core abilities of any autonomously navigating agent. While classical robotics usually estimates maps in a stand-alone manner with SLAM variants, which maintain a topological or metric representation, end-to-end learning of navigation keeps some form of memory in a neural network. Networks are typically imbued with inductive biases, which can range from vectorial representations to birds-eye metric tensors or topological structures. In this work, we propose to structure neural networks with two neural implicit representations, which are learned dynamically during each episode and map the content of the scene: (i) the Semantic Finder predicts the position of a previously seen queried object; (ii) the Occupancy and Exploration Implicit Representation encapsulates information about explored area and obstacles, and is queried with a novel global read mechanism which directly maps from function space to a usable embedding space. Both representations are leveraged by an agent trained with Reinforcement Learning (RL) and learned online during each episode. We evaluate the agent on Multi-Object Navigation and show the high impact of using neural implicit representations as a memory source.
Fichier principal
Vignette du fichier
ICCV2023_Implicit_Representations_Camera_Ready_arXiv.pdf (10.1 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-04390240 , version 1 (12-01-2024)

Licence

Identifiers

  • HAL Id : hal-04390240 , version 1

Cite

Pierre Marza, Laetitia Matignon, Olivier Simonin, Christian Wolf. Multi-Object Navigation with dynamically learned neural implicit representations. ICCV 2023 - International Conference on Computer Vision, Oct 2023, Paris, France. pp.1-21. ⟨hal-04390240⟩
37 View
16 Download

Share

Gmail Mastodon Facebook X LinkedIn More