Evidence Hub

A Hybrid Approach for FMCW Radar Signal Augmentation: DAPO and GSPO Integration

Qiaowang, Caiyun Wu

2025

Abstract

With the advancement of automated factories and intelligent logistics systems, AGVs (Automated Guided Vehicles) face increasing navigation challenges in complex outdoor environments such as grassy areas and gravel roads. LiDAR (Light Detection and Ranging) experiences a sharp drop in signal-to-noise ratio under adverse conditions such as rain, fog, and intense sunlight. In contrast, FMCW (Frequency-Modulated ContinuousWave) radar offers better anti-interference performance and lower cost. However, traditional signal processing methods often suffer from insufficient and diverse data for target detection. To address this, this paper presents a two-dimensional grassland environment dataset combining real FMCW radar signals and simulated data. The dataset combined actual FMCW radar signals and simulated data containing a total of$\mathbf{1 2, 0 0 0}$samples with the proportion of real signals to simulated ones being$3: 7$and the specific parameters for data collection including a detector radius ranging from$0.5-2.0$meters, a target velocity between$0.2-3.0 ~\mathrm{m} / \mathrm{s}$, and a vegetation density from$\mathbf{1 0 - 5 0}$plants per square meter, and the proposed pipeline increased the signal-to-noise ratio by 3.1 decibels and enhanced cosine similarity from 0.69 to 0.81 on the test set showing improved robustness against clutter and interference, indicating that this method had great potential for augmenting FMCW radar data and enhancing radar system performance in…

Read the paper

Related studies

Go deeper in Evidence Hub

Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.

Open this study Sign up free