Rescue Scout: RC Car Search-Rescue System

For my Project One, I built an RC-based Search and Rescue support system designed to help rescuers detect people in difficult or low-visibility environments. The idea is to use a simple RC car as a mobile camera platform while all detection and control are handled from a ground station.

The system uses a Raspberry Pi 5, live camera feeds, a Gradio touchscreen interface, and a YOLOv26 object detection model trained to recognize different human positions such as standing, sitting, lying, crawling, and partially hidden people. When a person is detected, the operator can immediately see the result on the interface and receive visual feedback through status LEDs.

This project is not meant to replace professional rescue equipment, but to explore how affordable hardware, computer vision, and a clear user interface can support search-and-rescue operations. The focus is on reliable person detection, simple operation, and a compact ground station that can be used in the field.

In this Instructable, I will explain the hardware, software, AI model, wiring, interface, and design choices behind the project.

Main Computing Unit

1. Raspberry Pi 5 - Link
2. 7-inch HDMI capacitive touchscreen - Link
3. MicroSD card for Raspberry Pi OS - Link

Camera and Detection System

1. RTSP Wi-Fi IP camera - Link
2. Camera mount or 3D printed holder - Self Designed (File uploaded)
3. Telescopic pole / selfie-stick support - Link
4. USB / power cables for the camera - Link
5. Power bank or external 5V power supply - Link

Mobile Platform

1. RC off-road car chassis - Link
2. Mounting hardware for attaching the camera system - Link

Electronics & Sensors

1. Green status LEDs - Link
2. RGB LED module - Link
3. Active buzzer - Link
4. MQ-2 gas sensor - Link
5. DHT11 temperature and humidity sensor - Link
6. PCF8591 ADC module - Link
7. Resistors for voltage divider circuits - Link
8. Jumper wires and Dupont connectors - Link
9. Prototype board / small PCB - Link
10. Battery for Raspberry Pi - Link

Mechanical Parts

1. ABS panels for the ground station enclosure - Link
2. Screws, nuts, spacers, and standoffs - Link
3. Double-sided tape / hot glue
4. Cable ties for cable management - Link
5. Cable Holders - Link

️ Tools

1. Soldering iron
2. Wire stripper / cutter
3. Screwdriver set
4. Drill or rotary tool
5. Multimeter
6. 3D printer or access to a 3D printing service
7. Laser cutter or hand tools for enclosure cutting

Software

1. Python
2. Gradio
3. OpenCV
4. Ultralytics YOLOv26
5. FastAPI
6. Docker / Docker Compose
7. Git and VS Code

Self-designed parts files:

Before building the system, I first defined the main goal of the project: to create a small Search and Rescue support platform that can help an operator detect people in difficult environments.

The RC car is used as a mobile camera platform. It is not autonomous and does not make rescue decisions by itself. Its purpose is to carry the camera into areas where it may be harder or unsafe for a person to quickly inspect.

The ground station is the main control point of the system. It contains the Raspberry Pi 5, the touchscreen interface, status LEDs, sensors, and the AI detection software. The operator can watch the live camera feed, see detection results, and receive clear visual feedback when a person is detected.

The system is built around four main parts:

Main Parts of the System

1. RC car platform — carries the camera system into the search area.
2. Camera feed — sends live video to the ground station.
3. Raspberry Pi 5 ground station — runs the interface and detection system.
4. YOLO person detection model — detects people in different positions, such as standing, sitting, lying, crawling, or partially hidden.

The main idea is to keep the system simple, affordable, and easy to understand. Instead of building a fully autonomous rescue robot, this project focuses on assisting the human operator with better visibility and AI-based person detection.

For the first version of the project, the system is designed as a prototype. It demonstrates how computer vision, embedded hardware, and a clear touchscreen interface can work together in a Search and Rescue scenario.

After defining the concept, I prepared the main hardware parts of the system and planned how they would work together.

The project is divided into two physical parts: the mobile RC car platform and the ground station. The RC car carries the camera, while the ground station contains the Raspberry Pi, touchscreen, sensors, buttons, LEDs, and detection software.

RC Car Platform

The RC car is used as a simple mobile base. Its main job is to move the camera into the search area. I did not modify it for autonomous driving, because the focus of this project is person detection and operator support.

For the camera setup, I used a small Wi-Fi / RTSP camera mounted on a pole or holder. This gives the camera a higher viewing angle and helps it detect people better in different positions.

Ground Station

The ground station is the control center of the project. It includes the Raspberry Pi 5, the touchscreen interface, status indicators, buttons, and sensors.

The Raspberry Pi runs the software, receives the live camera feed, and processes the video with the YOLO detection model. The touchscreen allows the operator to interact with the system in a simple way.

Components Prepared

For this step, I prepared:

1. Raspberry Pi 5
2. 7-inch HDMI touchscreen
3. RC off-road car
4. RTSP Wi-Fi camera
5. Camera mount / telescopic pole
6. Power supply or power bank
7. Status LEDs
8. RGB LED module
9. Physical buttons
10. MQ-2 gas sensor
11. DHT11 temperature and humidity sensor
12. PCF8591 ADC module
13. Jumper wires and connectors
14. ABS/acrylic panels for the enclosure

Before connecting everything, I checked the size and placement of the components. This helped me decide where the screen, buttons, LEDs, sensors, and cables would fit inside the ground station.

Planning the layout first is important, because the project contains both electronics and mechanical parts. A clean layout makes the final system easier to use, easier to repair, and more reliable during testing.

In this step, I prepared the RC car to work as the mobile camera platform for the Search and Rescue system.

The RC car itself is not autonomous. It is controlled manually and is mainly used to carry the camera into the search area. This keeps the project simpler and makes the focus clear: the main goal is not self-driving, but live video, person detection, and operator support.

Mounting the Camera

I mounted the camera on top of the RC car using a camera holder and a telescopic pole. The higher position gives the camera a better view of the environment and helps the AI model detect people more clearly.

The camera mount should be stable enough so the video feed does not shake too much while the car is moving. A shaky camera makes detection less reliable, especially when the person is far away or partially hidden.

Camera Placement

For the first version, I placed the camera so it looks forward from the RC car. This gives the operator a clear search direction and makes the system easier to control.

When placing the camera, I checked:

1. The camera has a clear forward view.
2. The camera is not blocked by the car body.
3. The mount is not too heavy for the RC car.
4. The camera cables or power cable do not touch the wheels.
5. The car can still drive and turn normally.

Power and Cable Management

The camera can be powered by its own battery or by a small power bank, depending on the camera type. I made sure the power cable was fixed to the frame so it would not disconnect while driving.

Cable management is important because loose wires can get caught in the wheels or make the car harder to control.

Result

After this step, the RC car became a simple mobile camera unit. It can drive into a search area and send a live video feed to the ground station, where the Raspberry Pi processes the image and runs person detection.

This part of the project acts like a “mobile tripod” for the camera, while the real intelligence of the system stays in the ground station.

In this step, I built the ground station for the Search and Rescue system. The ground station is the main control unit of the project. It receives the camera feed, runs the AI detection model, shows the interface, and gives feedback to the operator.

The ground station is built around a Raspberry Pi 5 and a 7-inch touchscreen. The Raspberry Pi runs the software, while the touchscreen is used as the main interface for the operator.

Main Purpose of the Ground Station

The ground station has several important jobs:

1. Show the live camera feed
2. Run the YOLO person detection model
3. Display detection results on the interface
4. Give visual feedback through LEDs
5. Allow the operator to control the system using buttons and the touchscreen
6. Hold extra sensors for environmental feedback

Component Placement

Before mounting the parts permanently, I planned the layout of the enclosure. I wanted the screen to be easy to see and the buttons to be easy to reach.

The touchscreen was placed on the front side of the enclosure. This makes it easy for the operator to view the camera feed and interact with the system.

The LEDs were placed near the front so the operator can quickly see the system status. The buttons were placed in an accessible position so they can be used without opening the enclosure.

The sensors were placed where they could still measure the environment properly and not be completely blocked inside the case.

Mounted Components

Inside and on the ground station, I mounted:

1. Raspberry Pi 5
2. 7-inch HDMI touchscreen
3. Status LEDs
4. RGB LED module
5. Physical push buttons
6. Active buzzer
7. MQ-2 gas sensor
8. DHT11 temperature and humidity sensor
9. PCF8591 ADC module
10. Power supply cables
11. Internal wiring and connectors

Enclosure Design

For the enclosure, I used flat panels made from ABS or acrylic-style material. The goal was to make the station compact, strong enough for testing, and easy to access if something needed to be repaired.

I also tried to keep the inside layout clean. Good cable management is important because the ground station contains many small electronic parts. Loose wires can cause connection problems, especially when the system is moved around during testing.

Result

After this step, the ground station became the main control center of the project. It holds the Raspberry Pi, screen, electronics, sensors, and user controls in one compact unit.

This makes the system easier to use because the operator does not need a laptop or separate monitor. Everything important is placed inside one portable ground station.

In this step, I connected the electronic components of the ground station to the Raspberry Pi 5.

Before wiring anything, I turned off the Raspberry Pi and disconnected the power. This is important because connecting sensors or LEDs while the Pi is powered can damage the GPIO pins.

The ground station uses several small electronic modules to give feedback to the operator and collect basic environmental data.

Connected Components

For this project, I connected:

1. Green status LEDs
2. RGB LED module
3. Physical push buttons
4. Active buzzer
5. MQ-2 gas sensor
6. DHT11 temperature and humidity sensor
7. PCF8591 ADC module

The LEDs are used to show system status, video feed status, and detection status. The buzzer can be used for alerts when a person is detected. The sensors provide extra environmental information for the operator.

Raspberry Pi GPIO Connections

The main GPIO connections used in my setup were:

ComponentRaspberry Pi Pin / GPIO

1. Active buzzer - GPIO 5
2. Green LED 1 - GPIO 23
3. Green LED 2 - GPIO 17
4. RGB LED Red - GPIO 26
5. RGB LED Green - GPIO 16
6. RGB LED Blue - GPIO 6
7. DHT11 sensor - GPIO 25
8. I2C SDA - GPIO 2
9. I2C SCL - GPIO 3
10. PCF8591 ADC - I2C address 0x4b

The PCF8591 ADC module is connected through I2C. This is needed because the Raspberry Pi cannot directly read analog signals. The MQ-2 gas sensor has an analog output, so it needs an ADC module to convert the signal into digital data that the Raspberry Pi can read.

MQ-2 Gas Sensor

The MQ-2 sensor was connected through the PCF8591 ADC module. The analog output from the MQ-2 goes to one of the analog input channels on the ADC.

A voltage divider was used where needed to keep the signal safe for the Raspberry Pi. This is important because Raspberry Pi GPIO pins use 3.3V logic and should not receive 5V directly.

I2C Wiring

The I2C bus uses two main wires:

1. SDA connected to GPIO 2
2. SCL connected to GPIO 3

Both the Raspberry Pi and the I2C module must also share the same ground connection. Without a common ground, the sensor readings may not work correctly.

Cable Management

After testing the connections, I organized the wires inside the enclosure. I tried to keep the cables short and separated where possible, especially around the screen, buttons, and sensors.

Good cable management makes the system easier to debug and reduces the chance of loose wires disconnecting during testing.

Result

After this step, the ground station electronics were connected to the Raspberry Pi. The LEDs, buzzer, sensors, and ADC module were ready to be tested with Python scripts before being integrated into the final interface.

Step 6: Set Up the Raspberry Pi Software

In this step, I prepared the Raspberry Pi 5 to run the ground station software.

The Raspberry Pi is the main computer of the system. It receives the camera feed, runs the AI detection model, controls the touchscreen interface, and communicates with the connected sensors, LEDs, buttons, and buzzer.

First, I installed Raspberry Pi OS on a microSD card and completed the basic setup. I configured Wi-Fi, SSH access, the touchscreen display, Python, and I2C support for the sensor modules.

After that, I updated the Raspberry Pi and installed the basic development tools:

I also created a Python virtual environment for the project and installed the required libraries, such as Gradio, OpenCV, Ultralytics YOLO, FastAPI, and GPIO-related packages.

The camera feed is read by the Raspberry Pi and shown inside the Gradio interface. The same video feed is also sent to the AI model for person detection.

The touchscreen interface was designed for the 7-inch display, so the ground station can be used without an external laptop or monitor.

After the setup was complete, I could start the project with:

When the application starts, the operator can open the interface on the Raspberry Pi touchscreen and see the live camera feed, detection results, and system information.

After this step, the Raspberry Pi was ready for testing the complete Search and Rescue system.

In this step, I added the AI model that detects people in the camera feed.

For this project, I used a YOLO object detection model. The model was trained to recognize people in different positions, such as standing, sitting, lying down, crawling, and partially hidden.

After training, I used the best model file in my Raspberry Pi software. The camera feed is sent to the model, and the model checks each frame for people.

When a person is detected, the result is shown on the touchscreen interface with a bounding box. The system can also use LEDs or alerts to make the detection easier to notice.

This makes the ground station more useful for Search and Rescue, because the operator can see both the live video and the AI detection result in one place.

In this step, I tested the live camera feed inside the ground station interface.

The camera sends video to the Raspberry Pi, and the Raspberry Pi shows it on the touchscreen. This allows the operator to see what the RC car camera is looking at in real time.

I checked that:

1. The camera connects correctly
2. The video appears on the touchscreen
3. The image is clear enough for detection
4. The feed is stable while the RC car is moving
5. The AI model can detect people from the live video

After the camera feed worked, I tested the system by placing people in different positions, such as standing, sitting, and lying down.

This step helped me check if the camera, Raspberry Pi, interface, and AI model were working together correctly.

In this step, I tested the full Search and Rescue system to make sure all parts worked together.

I started the Raspberry Pi ground station, opened the touchscreen interface, and connected the live camera feed. Then I tested the AI detection by placing people in different positions, such as standing, sitting, lying down, and partially hidden.

I checked that:

1. The camera feed appears correctly on the screen
2. The AI model detects people in the video
3. The bounding boxes are visible in the interface
4. The LEDs and alerts react when detection happens
5. The buttons and sensors work correctly
6. The system stays stable while running

This test helped me see if the RC car, camera, Raspberry Pi, interface, sensors, and AI model worked as one complete system.

After this step, the project was ready for final demonstration and improvements.

Step 10: Final Result, What I Learned and Future Improvements

After testing the complete system, the Search and Rescue prototype was ready for demonstration.

The final setup combines an RC car camera platform, a Raspberry Pi 5 ground station, a touchscreen interface, sensors, LEDs, and an AI person detection model. The operator can view the live camera feed, see detection results, and receive visual feedback when a person is detected.

This project shows how affordable hardware and AI can be used to support Search and Rescue tasks. The system is still a prototype, but it demonstrates the main idea clearly: helping an operator search an area with better visibility and automatic person detection.

During this project, I learned how to combine hardware, software, AI, and interface design into one working system. I also learned how important testing is, because every part of the project needs to work together: the camera, Raspberry Pi, sensors, touchscreen, AI model, and power system.

Future Improvements

In the future, the project could be improved by:

1. Making the camera feed more stable
2. Adding the Night Camera Vision or Heat detection
3. Improving detection of small or partially hidden people
4. Adding a stronger battery system
5. Making the enclosure more compact and durable
6. Testing the system in more realistic outdoor environments
7. Adding GPS location tracking + Virtual Detection Map interface
8. Improving the interface for faster use during emergencies

Overall, this project was a good way to build a practical Search and Rescue prototype using computer vision, embedded hardware, and a clear ground station interface.