Drone Perception and Localization system Proof of Concept

Block Diagram

Description

Nature Scanner is a weather-resistant, multi-sensor environmental sensing device developed as a proof-of-concept hardware platform for drone perception and localization. This device has potential uses in environmental sensing for agricultural applications like scanning microclimates or monitoring conditions inside storage facilities. I presented my project during Drone day 2025, an academic/industry UAV workshop hosted by the University of Arizona Agricultural Department in Tucson Arizona.

This device collects temperature, humidity, RGB images and GPS coordinates and stores them locally on an SD card. This suite of environmental and positional data is key for developing advanced drone perception and localization capabilities. The device consists of two submodules each controlled by an ESP32-CAM microcontroller development board. Both boards run custom firmware that allows the chips to communicate with sensor peripherals and can also send data wirelessly between boards. Sensor readings are managed by FreeRTOS,a real time operating system that allows NS to collect data to be collected continuously and reliably.

The “Josie” module utilizes an OmniVision OV2640 CMOS camera to capture 2 Megapixel (1600 x 1200 UXGA) images set to one frame per second. It was specifically designed to save these high-quality, uncompressed images directly onto the onboard micro SD card for later retrieval and analysis. The “Adam” module uses a Sensirion SHT31 solid state temperature/humidity sensor alongside a HiLetgo GY-NEO6MV2 NEO-6M GPS. The NEO6 GPS receives a satellite signal at L1 band 1575.42 MHz. GPRMC sentences are logged in a text file with UTC time, position status, lat, lon, speed (knots), date and more. Temperature and relative humidity (at resolution of 0.01 °C/%RH) are also stored in the same text file with a timestamp.

Software and Firmware: Architecture, Implementation, and Technical Challenges

At the highest level, nature scanner consists of two versions of custom firmware that live on two ESP32-cam microcontrollers. I used Espressifs ESP IDF SDK to adapt prebuilt drivers to initialize and control sensors as well as communicate between dev boards. FreeRTOS xTaskCreate was used to schedule telemetry, commanding and data flow processes enabling simultaneous management of resources.

Each firmware application follows standard C practices with initialization, task definitions, and entry points in main.c. Leveraging the ESP-IDF build system, function definitions and prototypes are organized into component folders. This modular structure enhances maintainability and allows for easier reconfiguration of sensor assignments between the two boards.

The primary motivation for the two-controller architecture stemmed from hardware limitations: GPIO ports required to communicate with peripherals over I2C/UART conflicted with SD card interface pins when operating in 4-bit SDMMC mode. The ESP32's SD/MMC host peripheral pins are hardwired to the physical SD card slot, leaving insufficient GPIOs for other components on a single board. Attempting a single-board design resulted in DMA and write block errors during simultaneous SD card access and peripheral operation.

Communication protocols include I2C for the SHT31 temp sensor and UART for the NEO6M GPS module. The ESP32 interfaces with the OV2640 camera via SPI to read image data, which is subsequently written to the SD card using the SDMMC protocol.

Writing the SHT31 temp sensor driver initially proved challenging. The suggested driver within the framework (i2c.c) was an older, procedural library focused on sequential I2C operations (create, start, write). Instead, I utilized the ESP-IDF (i2c_master driver), which is designed around bus/device handles and allows for complete transactions. Utilizing functions like i2c_master_transmit() within this driver, I successfully implemented commands specified from the Sensirion user manual, such as 'Periodic Data Acquisition Mode', to continuously retrieve sensor data.

Hardware BOM

Category	Component	Description	Approximate Cost (USD)	Purchase
microcontroller	ESP32-CAM dev board x2	dual-core 32-bit LX6 CPU, 2MP OV2640 cam module, microSD slot, wifi, bluetooth	$14.00	Purchased
Sensor	SHT31-DIS-B Sensor	High-precision humidity and temperature sensor, I2C interface	$12.88	Purchased
	Ublox NEO-M6 GPS Module	GPS module with 1-2 meter accuracy, EEPROM supports multiple satellite systems	$9.00	Purchased
Encloser	ABS Plastic IP65 Enclosure	Waterproof, dustproof project enclosure (5.9 x 3.9 x 2.8 inch)	$13.99	Purchased
serial interface	HiLetgo Serial adapter	FT232RL Mini USB to TTL Serial Converter Adapter Module	$6.49	Purchased
memory	SD card	SanDisk 32GB 2-Pack Ultra MicroSDHC UHS-1 Memory Card (2x32GB)	$13.56	Purchased
PCB	PCB	Prototype PCB Solderable Breadboard(5 Pack + 1 Mini Board, Red)	$8.49	Purchased
Power	voltage regulator	AMS1117-3.3V Buck Module LDO 800MA	$0.60	Purchased
	voltage regulator	5v Regulator Module Mini Voltage Reducer DC 4.5-24V 12V 24V to 5V 3A	$1.10	Purchased
	battery	7.4V LiPo Battery 600mAh 2S 30C Rechargeable Lithium Polymer Batteries	$17.00	Purchased

Development Environment

Category	Tool	Purpose / Notes
Framework	ESP-IDF	Espressif IoT Dev Framework for ESP32-CAM
	CMake, Ninja	Build system tools used by ESP-IDF
	menuconfig	ESP-IDF tool for project configuration
RTOS	FreeRTOS	Real-Time Operating System (integrated in ESP-IDF)
Language	C	Primary firmware development language
Compiler	Xtensa GCC Toolchain	C/C++ compiler for ESP32 Xtensa core
IDE	Visual Studio Code (VS Code)	Integrated Development Environment
	ESP-IDF VS Code Extension	Integrates ESP-IDF build, flash, debug in VS Code