Frequency modulated continuous wave (FMCW) radar is increasingly used for various detection and classification applications in different fields, such as autonomous vehicles and mining fields. Our objective is to increase the classification accuracy of objects detected using millimeter-wave radar. We have developed an approach based on millimeter-wave radar. The proposed solution combines the use of an FMCW radar, a YOLOv7 model, and the Pix2Pix architecture. The latter architecture was used to reduce noise in the heatmaps. We create a dataset of 4125 heatmaps annotated with five different object classes. To evaluate the proposed approach, 14 different models were trained using the annotated heatmap dataset. In the initial experiment, we compared the models using metrics such as mean average precision (mAP), precision, and recall. The results showed that the proposed model of YOLOv7 (YOLOv7-PM) was the most efficient in terms of mAP_0.5, which reached 90.1%, and achieved a mAP_0.5:0.95 of 49.51%. In the second experiment, we compared the models with a cleaned dataset generated using the Pix2Pix architecture. As a result, we observed improved performances, with the Pix2Pix + YOLOv7-PM model achieving the best mAP_0.5, reaching 91.82%, and a mAP_0.5:0.95 of 52.59%.

Recently, mmWave radars have been used in several automotive and industrial applications to provide the position and speed of detected objects using FMCW algorithms. However, this radar is not able to recognize or classify the detected objects. This paper associates a mmWave radar with a real-time embedded computing platform and proposes a real-time implementation of deep learning algorithms in order to classify the detected objects. Three different classes of objects have been studied in this paper, i.e., human, car, and motorcycle. The proposed on-board system is based on the AWR2944EVM as a mmWave radar and the Nvidia Jetson Xavier NX as an embedded computing platform. The system was mounted on a motorcycle for data acquisition and real-time classification. The system developed captures radar data and processes it rapidly, converting it into usable formats for efficient object classification. Several YOLO detection deep learning models were evaluated, among which the YOLOv9-E model, which provided the best accuracy and speed performances for our application, i.e., a mAP_0.5 of 84% and an accuracy of 86.4%. The proposed architecture can process up to 18.59 frames per second (FPS). This system has been tested in a wide range of real-life conditions, validating its robustness and ability to operate in a variety of environments. The results obtained show that the proposed implementation delivers good real-time inference performance with optimized power consumption.

Traffic sign (TS) detection and recognition are essential components of advanced driver assistance systems (ADAS), contributing to safer and more reliable driving. However, deploying deep learning–based vision models on embedded platforms is challenging due to constraints in computational power and energy consumption. In this work, a comparative deployment of you only look once version 7 (YOLOv7) and YOLOv7-tiny deep learning algorithms is conducted on embedded NVIDIA platforms, namely Jetson Nano and Jetson Xavier NX, to evaluate their suitability for real-time TS detection. Following the detection stage, a convolutional neural network (CNN) is integrated to perform TS recognition, enabling a complete detection–recognition pipeline. Experimental results show that YOLOv7-tiny achieves higher detection precision of 97%, while providing better speed and computational cost on resource-constrained devices, with Jetson Nano reaching 18.8 frames per second (FPS) and, on Jetson Xavier NX reaching 43 FPS. The integrated CNN model ensures reliable classification of detected TS with an accuracy of 99.54%. This work highlights the trade-offs between precision, speed, and power consumption and provides practical guidance for selecting detection and recognition architectures for embedded ADAS applications.

This article describes the design, control, and implementation of a photovoltaic (PV) emulator using two-switch forward-converter topology. The system is designed to emulate the nonlinear electrical behavior of an actual PV panel under different environmental conditions including radiation level and temperature. The emulator provides galvanic isolation and also accurate current modulation to provide a safe yet reliable means of testing PV-related devices and algorithms within a laboratory setting. A dual-loop PI control is proposed to adjust the output current according to voltage feedback (VF), thus making accurate I–V and P–V curves achievable. Besides software simulation, a tailored printed circuit board (PCB) was fabricated. The simulation result demonstrated that the system can achieve a fast response and stable operation, with a maximum error percentage of about 2.1%, indicating high emulation fidelity, thereby providing an attractive platform for various evaluation purposes such as MPPT algorithms, inverters, and EMS.

!Traffic sign (TS) detection is a key component of driver assistance systems in intelligent automotive applications, as it enhances road safety and helps prevent accidents. However, several challenges can affect detection performances, including partially occluded signs, fluctuating detection confidence, damaged signs, adverse weather conditions, and varying lighting environments. Moreover, deploying such systems on resource-constrained embedded platforms presents significant challenges due to limited computational power, memory constraints, and strict real-time requirements. To address these challenges, this paper proposes the use of an optimized approach for real-time TS detection using YOLOv8 deployed on embedded platforms. The proposed system is designed to address the constraints of edge devices, including limited computational resources and power consumption, while maintaining high detection performance. Different variants of YOLOv8 have been evaluated in terms of accuracy and real-time processing constraints. After detecting the TS categories, a convolutional neural network (CNN) model is proposed for TS recognition to determine the exact indication of each TS. The experimental results show that the best detection model achieves a mean Average Precision (mAP@0.5) of 99% across the five TS categories, while the recognition model reaches an accuracy of 99.54%. Moreover, a proposed implementation of the detection model has achieved approximately 92 frames per second (FPS) running on the Jetson Orin Nano embedded platform, which is suitable for Advanced Driver Assistance Systems (ADAS) and intelligent transportation applications.