Can NeRFs See without Cameras?

2025-05-28

AI Summary

This work addresses implicit geometric perception in camera-free indoor environments by reconstructing floorplans from sparse WiFi received signal strength (RSS) measurements. We propose RF-NeRF, the first Neural Radiance Fields (NeRF) framework adapted to the radio-frequency (RF) domain: it integrates a physics-informed multipath propagation model with differentiable ray tracing, using RSS as the sole supervisory signal to learn an implicit scene representation end-to-end. Unlike prior approaches, RF-NeRF requires neither prior maps nor simultaneous localization and mapping (SLAM); it reconstructs structurally accurate 2D floorplans from only a small number of WiFi RSS samples collected at known locations within a residence. Experiments demonstrate that the reconstructed geometry enables high-fidelity wireless signal prediction and basic ray tracing, validating the feasibility of non-optical NeRF for RF sensing. This establishes a new paradigm for privacy-preserving, low-cost indoor modeling.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Planning, Routing, and Scheduling: Activity and Plan Recognition
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Location- and context-aware Web and WoT applications and services
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Neural Radiance Fields (NeRFs) have been remarkably successful at synthesizing novel views of 3D scenes by optimizing a volumetric scene function. This scene function models how optical rays bring color information from a 3D object to the camera pixels. Radio frequency (RF) or audio signals can also be viewed as a vehicle for delivering information about the environment to a sensor. However, unlike camera pixels, an RF/audio sensor receives a mixture of signals that contain many environmental reflections (also called"multipath"). Is it still possible to infer the environment using such multipath signals? We show that with redesign, NeRFs can be taught to learn from multipath signals, and thereby"see"the environment. As a grounding application, we aim to infer the indoor floorplan of a home from sparse WiFi measurements made at multiple locations inside the home. Although a difficult inverse problem, our implicitly learnt floorplans look promising, and enables forward applications, such as indoor signal prediction and basic ray tracing.

Problem

• Can NeRFs learn from multipath signals like RF/audio?
• Infer indoor floorplans from sparse WiFi measurements.
• Enable applications like signal prediction and ray tracing.

Innovation

• NeRFs learn from multipath RF/audio signals
• Infer indoor floorplans from WiFi measurements
• Enable indoor signal prediction and ray tracing