The increasing demand for fast and cost effective last mile delivery solutions has catalyzed significant advancements in drone based logistics. This research describes the development of an AI integrated drone delivery system, focusing on route optimization, object detection, secure package handling, and real time tracking. The proposed system leverages YOLOv4 Tiny for object detection, the NEO 6M GPS module for navigation, and the A7670 SIM module for real time communication. A comparative analysis of lightweight AI models and hardware components is conducted to determine the optimal configuration for real time UAV based delivery. Key challenges including battery efficiency, regulatory compliance, and security considerations are addressed through the integration of machine learning techniques, IoT devices, and encryption protocols. Preliminary studies demonstrate improvement in delivery time compared to conventional ground based logistics, along with high accuracy recipient authentication through facial recognition. The study also discusses ethical implications and societal acceptance of drone deliveries, ensuring compliance with FAA, EASA and DGCA regulatory standards. Note: This paper presents the architecture, design, and preliminary simulation results of the proposed system. Experimental results, simulation benchmarks, and deployment statistics are currently being acquired. A comprehensive analysis will be included in the extended version of this work.

The ESP32-CAM is one of the most widely adopted open-source modules for prototyping embedded vision applications. Since its release in 2019, it has gained popularity among both hobbyists and professional developers due to its affordability, versatility, and integrated wireless capabilities. Despite its widespread use, comprehensive documentation of the performance metrics remains limited. This study addresses this gap by collecting and analyzing over six hours of real-time video streaming logs across all supported resolutions of the OV2640 image sensor, tested under five distinct voltage conditions via an HTTP-based WiFi connection. A long standing bug in the official Arduino ESP32 driver, responsible for inaccurate frame rate logging, was fixed. The resulting analysis includes key performance metrics such as instantaneous and average frame rate, total streamed data, transmission count, and internal chip temperature. The influence of varying power levels was evaluated to assess the reliability of the module.

This article is a quick getting started guide for the ESP32-CAM board. Note: in this tutorial we use the example from the arduino-esp32 library. This tutorial doesn't cover how to modify the example. You can watch the video tutorial or keep reading this page for the written instructions. You can use the preceding links or go directly to MakerAdvisor.com/tools to find all the parts for your projects at the best price!

This tutorial covers how to use TinyML with ESP32-CAM. It describes how to classify images using ESP32-CAM with a machine learning model running directly on the device. To do it, it is necessary to create a machine learning model using Tensorflow lite and shrink the model. There are several ways to do it, this tutorial uses Edge Impulse that simplifies all the steps. We will explore the power of TinyML with ESP32-CAM to recognize and classify images.