Heaven’s Light is Our Guide
Department of Electrical & Electronic Engineering
Rajshahi University of Engineering & Technology
Rajshahi-6204, Bangladesh
Proforma for submitting Research Project
Course Code: EEE 3100 Course Title: Electronic Shop Practice
1. Name and Roll No. :
2. Name of the Department : Electrical & Electronic Engineering
3. Name of the University : Rajshahi University of Engineering &
Technology
4. Title of the Project : AI-Based GPS-Free Smart Object
Tracking System Using Radar and
ESP32-CAMs
5. Brief description of the project:
(a) Objectives and aims of the project (within 100 words): The objective of this project is to
design a smart surveillance and tracking system that can detect, identify, and track people,
vehicles, or animals across public areas using only radar and camera-based AI — without
using any GPS or physical tracking device. The system assigns a soft digital ID to each
detected object and follows them across zones using object detection, motion data, and
appearance-based re-identification.
(b) Review of literature on the subject matter of the project and rationale behind the present
initiative (within 500 words): Traditional object tracking systems often rely on GPS
trackers, which are impractical for public surveillance since not all individuals or vehicles
carry one. Modern surveillance technologies now use AI vision (YOLOv8) and radar
systems to track moving objects. Cameras such as the ESP32-CAM, paired with radar
sensors like the Hi-Link LD2450, offer a low-cost and scalable approach to building GPS-
independent tracking systems.
This project is inspired by smart city surveillance systems that use multi-camera AI and
radar fusion for motion detection and prediction. While large-scale systems often require
dense deployment, this project minimizes cost by applying predictive algorithms and
appearance-based re-identification (ReID) to track objects across larger gaps (500m–1km)
between nodes.
The LD2450 radar supports multi-target tracking and outputs real-time X/Y coordinates
and speed of up to three targets. Combined with ESP32-CAM's AI-based detection, the
system assigns soft IDs using Deep SORT and keeps track of movement even when objects
are temporarily hidden. The vision and radar data are processed by a central Python-based
Streamlit dashboard that simulates GPS movement using Folium maps.
The proposed design thus bridges the gap between high-cost GPS systems and blind, static
CCTV systems. It provides intelligent tracking using edge-level devices with potential
applications in traffic surveillance, perimeter security, and urban monitoring — especially
in budget-constrained environments.
(c) Expected results:
• Detection and tracking of people and cars using ESP32-CAM and YOLOv8
• Real-time motion data collection using LD2450 radar sensor
• Soft ID generation and re-identification across nodes
• Map-based GPS simulation for object movement
• Live dashboard with video, radar data, and logs
(d) Relevance of the project to national development:
This system enables low-cost, GPS-free object monitoring and surveillance, making it
ideal for smart city development, border monitoring, rural traffic management, and
public safety. It reduces infrastructure cost while enhancing real-time object awareness
in critical zones — aligning with national goals for digital transformation and secure
urban planning.
(e) Methodology to be adopted in the investigation (detail as far as possible)
Data collection:
• Object detection data from ESP32-CAM (via YOLOv8)
• Motion and distance data from LD2450 radar (via UART to PC)
Experimental work:
• Arduino + Python code to read and fuse radar-camera data
• ReID module for matching objects across zones
• Visualization using Streamlit and Folium
6. Financial Statement (item wise):
Serial Name of the Specification Quantity Unit Price Total Price
No. Equipment’s (Taka) (Taka)
1 ESP32-CAM MB Wi-Fi Camera 2 1495 2990
(with casing) with Mounting
Accessories
2 LD2450 Radar 24GHz radar (3 1 2950
2950
target tracking)
3 Arduino UNO R3 Microcontroller 1 890 890
Page 2 of 7
4 FTDI USB Adapter For ESP32- 1 200 200
CAM
programming
Jumper Wires + Connecting 1 set 150 150
Breadboard components
Total expense (Taka): 7180
Signature of the Course Teacher Signature of the Students
Name:
Date:
Email:
Signature of the Head of the Department
Page 3 of 7
DECLARATION/CERTIFICATION
It is certified that-----
a) The same project has not been submitted to anywhere (UGC/other agency/University) for financial
support.
(b) The research work proposed in this project is not a duplicate work already done or being done in
the field (i.e. area of research)
(c) We agree to accept the terms and conditions developed for the Allocation for Engineering, Science
and Technology as mentioned in the Guidelines.
(d) Project will be provided with access to all available facilities in this university.
Signature of the Course Teacher Signature of the Students
Name:
Date:
Email:
Signature of the Head of the Department
Page 4 of 7
Department of Electrical & Electronic Engineering
Rajshahi University of Engineering & Technology
Project Title : AI-Based GPS-Free Smart Object Tracking System Using
Radar and ESP32-CAMs
Project Description : This project aims to develop a smart object tracking system
that works without GPS by combining radar and AI-powered
camera modules. Using the ESP32-CAM for real-time object
detection and the LD2450 radar for motion tracking, the
system identifies and follows people or vehicles across
different zones.
Each detected object is assigned a digital "soft ID" and tracked
using both visual and radar data, even if temporarily out of
view. A central dashboard (Streamlit + Folium) displays live
camera feeds and simulates object movement on a GPS-style
map.
The solution is designed to be low-cost, scalable, and suitable
for GPS-denied environments like indoor areas, rural zones,
or disaster sites. It supports applications in public surveillance,
traffic monitoring, and smart city infrastructure.
Team Members (Name and : 1. Md. Al Shahriar Hossen Kabir(2101103)
Roll No) 2. Al-Momin Hosen(2101102)
3. Abdullah Al Rifat(2101092)
Technical Design and Implementation
This project uses the ESP32-CAM module and LD2450 24GHz radar sensor as the main
components. The ESP32-CAM is a low-cost board with built-in Wi-Fi, Bluetooth, and camera,
capable of running lightweight AI models like YOLOv8-tiny for real-time object detection. The
LD2450 radar provides X-Y coordinate and motion data for up to three moving targets via UART.
Other supporting ICs include:
• AMS1117 voltage regulator for 3.3V supply,
• FTDI adapter for ESP32-CAM programming,
• Logic level shifters for voltage compatibility,
• Optional GPIO expanders if more I/O pins are needed.
Functionality:
When motion is detected by the radar, the ESP32-CAM captures images and runs object detection.
Each object is assigned a soft ID using a ReID algorithm (like Deep SORT) and tracked across
Page 5 of 7
zones. Data is sent to a Python-based Streamlit dashboard, which shows object movement on a
simulated GPS map using Folium.
Circuit and PCB Design:
The circuit will be designed in EasyEDA, following standard rules:
• Minimum trace width: 0.25 mm; clearance: 0.3 mm,
• Use of ground plane for noise reduction,
• Decoupling capacitors near ICs,
• Separate power lines for radar and ESP32-CAM.
A 2-layer PCB will be designed, simulated, and fabricated. Pin headers will be added for easy
testing and modular expansion.
Project Justification and Value
This project solves the problem of object tracking in GPS-restricted areas by using radar and AI-
powered ESP32-CAM modules. It offers a low-cost, scalable solution for public surveillance
without needing physical trackers or GPS.
Value Added:
• Enables tracking in GPS-denied zones
• Reduces cost using affordable edge devices
• Enhances public safety and smart city infrastructure
Development Stages:
1. Literature review and system planning
2. Selection of ESP32-CAM and LD2450 radar
3. PCB and circuit design
4. AI integration (YOLOv8, Deep SORT)
5. Testing and dashboard visualization using Python
Justification:
ESP32-CAM and LD2450 offer an efficient blend of vision and motion tracking. The design is
modular, cost-effective, and suitable for real-time monitoring in both rural and urban areas.
Project Planning and Team Roles
Our team has divided the project into clear stages: system research, hardware setup, software
development, integration, and testing. Tasks are assigned based on individual strengths—for
example, one member focuses on AI and coding, another handles hardware and PCB design, and
another manages testing and documentation.
Economic factors influenced the use of low-cost modules like ESP32-CAM and LD2450 radar to
keep the project affordable. Technically, we prioritized components that are easy to integrate and
support open-source development.
We coordinate through regular group meetings, shared task tracking (Google Docs). This
structured planning ensures active participation and efficient project completion.
Page 6 of 7
(Addresses Complex Engineering Problem criteria)
This project tackles the challenge of object tracking without GPS, especially in areas where
GPS signals are unavailable or unreliable. It requires integrating radar sensing, AI-based vision,
and real-time data processing on low-power devices.
Complexity:
• Multidisciplinary integration: Embedded systems, radar, AI, and software
• System-level challenges: Sensor fusion, soft ID tracking, real-time performance
• Constraints: Limited hardware resources, power efficiency, and data accuracy
Solution Approach:
We apply systematic engineering by using modular hardware design, lightweight AI models, and
efficient data fusion methods to ensure reliable and scalable tracking without GPS.
Page 7 of 7