Noise Analysis and Modeling of the PMD Flexx2 Depth Camera for Robotic Applications

TL;DR

This work addresses the problem of accurate depth sensor simulation for PMD Flexx2 ToF cameras by characterizing non-systematic noise. It separates axial and lateral noise, modeling axial noise as a distance- and incidence-angle-dependent Gaussian with a fitted formula and mode-specific coefficients, and conservatively approximates lateral noise as Gaussian with distance/angle independence. The results show a low average KL divergence for axial noise ($0.015$ nats) and a higher but acceptable divergence for lateral noise ($0.868$ nats), supporting realistic sensor emulation. The findings facilitate improved sim-to-real transfer for exteroceptive RL controllers in mobile robotics, while future work will broaden surface material diversity to further enhance robustness.

Abstract

Time of Flight ToF cameras renowned for their ability to capture realtime 3D information have become indispensable for agile mobile robotics These cameras utilize light signals to accurately measure distances enabling robots to navigate complex environments with precision Innovative depth cameras characterized by their compact size and lightweight design such as the recently released PMD Flexx2 are particularly suited for mobile robots Capable of achieving high frame rates while capturing depth information this innovative sensor is suitable for tasks such as robot navigation and terrain mapping Operating on the ToF measurement principle the sensor offers multiple benefits over classic stereobased depth cameras However the depth images produced by the camera are subject to noise from multiple sources complicating their simulation This paper proposes an accurate quantification and modeling of the nonsystematic noise of the PMD Flexx2 We propose models for both axial and lateral noise across various camera modes assuming Gaussian distributions Axial noise modeled as a function of distance and incidence angle demonstrated a low average KullbackLeibler KL divergence of 0015 nats reflecting precise noise characterization Lateral noise deviating from a Gaussian distribution was modeled conservatively yielding a satisfactory KL divergence of 0868 nats These results validate our noise models crucial for accurately simulating sensor behavior in virtual environments and reducing the simtoreal gap in learningbased control approaches

Figures (7)

• Figure 1: Quadrupedal robot equipped with two PMD Flexx2 cameras.
• Figure 2: Testing environment setup, where $\Theta$ is the incident angle.
• Figure 3: \ref{['fig:axial_noise']}: A top-down cross-section of the depth map, \ref{['fig:lateral_noise']}: 2D projected depth map of a planar target.
• Figure 4: Examples of depth images with region chosen for axial noise analysis.
• Figure 5: Examples of depth images with region chosen for lateral noise analysis.