Edge Radar Material Classification Under Geometry Shifts

2026-03-24

AI Summary

This work addresses the sensitivity of millimeter-wave radar to sensor geometric pose variations—such as height and tilt angle—when performing material classification on ultra-low-power edge devices. To enable real-time inference under stringent resource constraints, the authors propose a lightweight classification method based on range-bin intensity features and a multilayer perceptron (MLP). They systematically model the feature shifts induced by geometric perturbations and investigate the underlying mechanisms through normalization, geometric data augmentation, and motion-aware feature design. Experiments conducted with a TI IWRL6432 radar demonstrate a macro-F1 score of 94.2% under nominal conditions, which degrades to 68.5% under realistic geometric deviations, thereby highlighting critical pathways and practical strategies for enhancing robustness in real-world deployments.

Technology Category

• Machine Learning: Hardware-aware ML
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Material awareness can improve robotic navigation and interaction, particularly in conditions where cameras and LiDAR degrade. We present a lightweight mmWave radar material classification pipeline designed for ultra-low-power edge devices (TI IWRL6432), using compact range-bin intensity descriptors and a Multilayer Perceptron (MLP) for real-time inference. While the classifier reaches a macro-F1 of 94.2\% under the nominal training geometry, we observe a pronounced performance drop under realistic geometry shifts, including sensor height changes and small tilt angles. These perturbations induce systematic intensity scaling and angle-dependent radar cross section (RCS) effects, pushing features out of distribution and reducing macro-F1 to around 68.5\%. We analyze these failure modes and outline practical directions for improving robustness with normalization, geometry augmentation, and motion-aware features.

Problem

• mmWave radar
• material classification
• geometry shifts
• edge devices
• radar cross section

Innovation

• mmWave radar
• material classification
• geometry shifts
• edge computing
• robustness