STEP-BY-STEP TUTORIAL · PHYSICAL COMPUTING & PHYSICAL AI LAB

How to Set Up a Physical Computing and Physical AI Lab with VTM Studio

By Tishin Padilla · VideotronicMaker Tutorials
When you are learning how to set up a physical computing lab and how to set up a physical AI lab at home, you do not need to buy a room full of equipment. You start with the computer and microcontroller you may already own, use VTM Studio as all-in-one physical computing software to plan and test your circuit on a breadboard, mount your parts onto a P³ (Physical Prototyping Platform) frame like a wooden milk crate, and add AI vision (using your laptop webcam or your cell phone with an app like Iriun), voice, and servo control as your build takes shape.

Step 1: Plan and Test Your Circuit on the Breadboard (From a Single Sensor to Remote LoRa Sensors)

Illustrated LoRa college campus project showing multiple remote sensors (water level sensor at the beach, photoresistor by the satellite dish, soil moisture sensor in the woods, temperature and humidity sensor on a building roof, and a Waveshare TSL2591 light sensor under a tree) paired with Arduino Nanos and Reyax LoRa transceivers transmitting back to a dorm room base station via an RYLS135 UART bridge.
Remote Campus Research with LoRa: Placing multiple remote sensors (water level, photoresistor, soil moisture, temperature/humidity, and a Waveshare TSL2591 light sensor) with Arduino Nanos and Reyax LoRa transmitter modules around a college campus, streaming long-range telemetry back to a dorm room receiver connected via a RYLS135 USB-to-UART bridge.

Most physical AI projects start out as physical computing projects on a breadboard. Imagine you are studying science, engineering, biology, or environmental studies on a college campus and you want to run a real field research project right from your dorm room on the far end of campus.

As shown in the campus map from my LoRa video above, say you have permission to place five remote sensors miles across campus where Bluetooth and Wi-Fi cannot reach and where you cannot run a wire:

  • A water level sensor down at the waterfront or beach
  • A photoresistor out by the campus satellite dish
  • A soil moisture sensor back in the woods
  • A temperature and humidity sensor on the roof of a campus building
  • A Waveshare `TSL2591` (`TSL25911`) precision light and infrared sensor under a tree

Each remote sensor is wired to a small microcontroller (like an Arduino Nano powered by a portable USB power bank) and a Reyax LoRa transceiver module acting as a transmitter. Because LoRa is a low-power, long-range wireless transceiver protocol, it can send small sensor packets (LUX:, TEMP:, SOIL:) miles through trees and buildings back to a second LoRa module in your dorm room—and because both modules are transceivers, you can even send a LoRa command back to turn a remote sensor on for 10 minutes once a day and shut it back off without Wi-Fi or Bluetooth.

Back in your dorm room (or at your home desk), the receiving LoRa module connects to a UART bridge / level shifter (`RYLS135`) and plugs directly into your laptop or a USB hub on your Arduino UNO Q. As you cover and uncover the TSL2591 light sensor—or walk away past walls and metal barriers—you see both the live lux levels (`LUX:`) and the LoRa signal strength (`RSSI:`) updating in real time on the chest screen of your desk bot or P³ workstation.

Before taking those remote LoRa sensors out into the field or mounting hardware onto a frame, you plan and test the circuit on your breadboard. (If you want to check circuit math before wiring physical parts, you can also start free in your browser with VTM Studio Lite using the circuit simulators and electronics calculators on VideotronicMaker.)

Illustrated 2D breadboard circuit layout for the remote LoRa transmitter showing an Arduino Nano, a Waveshare TSL2591 I2C light sensor connected to A4 (SDA) and A5 (SCL), and a Reyax RYLR998 LoRa transceiver module powered from 3.3 V with TXD on D2 and a resistor divider protecting the 3.3 V RXD line.
Remote LoRa Transmitter Breadboard Layout: An Arduino Nano wired to a Waveshare `TSL2591` light sensor over I2C (SDA to A4, SCL to A5) and a Reyax `RYLR998` LoRa transceiver powered from the 3.3 V rail (TXD to D2, with a voltage-divider resistor to ground protecting the 3.3 V LoRa RXD line).

Inside VTM Studio on your Mac, Windows PC, Linux desktop, Raspberry Pi 4B, or Arduino UNO Q, you work through five connected steps on the breadboard:

  1. Plan the Breadboard in Layout Mode & Spec Mode: Draw your 2D breadboard and circuit wiring in Layout Mode (saving .vtml project files or exporting layered .psd, .svg, and .png diagrams) while checking pinouts, schematics, and logic levels (like wiring the 3.3 V LoRa module and RYLS135 UART bridge) in Spec Mode.
  2. Check Resistor and Voltage Divider Math in Reference Mode: Use the calculators in Reference Mode to size a voltage divider (such as dropping a 5 V Arduino Nano D3 transmit pin down to safe 3.3 V logic with a 330 Ω resistor divider before it hits the LoRa module's RX line).
  3. Probe Your Breadboard in Meter Mode: Flash meter_sketch.ino to an Arduino and follow the 2-wire probe diagram in Meter Mode to measure low-voltage DC Voltage, Resistance (Ω), and Continuity on the backlit digital LCD.
  4. Flash Firmware in Code Mode: Open Code Mode, load one of the built-in Science Mode Examples (or your Nano + TSL2591 LoRa sketch) into the code editor, compile and flash your board over USB, and click [CONNECT TELEMETRY].
  5. Watch Live Gauges and `RAW:` Serial in Science Mode: Add the matching panel (LUX:, RSSI:, TEMP:, SOIL:, DIST:, or ANGLE:) in Science Mode to watch live lux levels and LoRa signal strength update as you walk through different rooms or cover the sensor, while verifying the unparsed RAW: serial string underneath.

Step 2: Move from the Plan on Screen to Mounting on Your P³ Frame

Getting the circuit working on a flat breadboard is only half the fun. Those five steps in VTM Studio are meant to lead directly to the next step: taking your working components and mounting them onto your frame.

Note that the LoRa remote sensor project shown in Step 1 and the robotic arm images shown below are two different projects—both start as physical computing projects and grow into physical AI projects as you add vision, voice, and AI reasoning in VTM Studio. They are here as examples of the different variations you can build with your P³ builds (though you can also combine them together so a single P³ station both receives remote LoRa telemetry and drives a physical arm).

Ideally, your plan is geared toward a useful build that makes sense for Physical AI—and it does not have to be a robot. Even if you do not have a final end-product in mind yet, the good news is that this is a P³ build—which stands for Physical Prototyping Platform. That means your first builds can simply be for testing, learning, and trying things out: figuring out how to mount an arm, where to mount your sensor, or how to keep wires out of the way of moving parts.

Close-up inside the wooden crate P3 frame showing the breadboard and glowing red Arduino Nano mounted vertically to the wooden wall with USB cable routed through the slats, and a MacBook Pro in the foreground running VTM Studio with Mouth Animation and MG90S 5-DoF Arduino PWM Servo Override sliders.
Mounting the Breadboard & Nano to the P³ Frame: In this robotic arm variation, the solderless breadboard with the Arduino Nano is mounted directly to the inside wooden wall of the crate with its USB cable routed through the slats, while VTM Studio runs in the foreground showing the Mouth Animation display and the ARMS (`MG90S 5-DoF Arduino PWM`) joint sliders.

Once you start building physically, you immediately start figuring out real-world logistics—because the moment you go from a plan on paper (or a 2D layout on screen) to the actual physical build, things change. You might think an arm belongs on the outside wall of the crate, and then realize while bolting it together that mounting it underneath the top slats hanging down gives it better reach or balance. You might move a microcontroller board to an inside slat so the USB cable reaches cleanly, or add a wooden dowel strut to brace a heavy joint. Figuring out those physical logistics hands-on is what makes building a P³ fun and exciting.

Close-up of the custom underside swivel mount inside the wooden crate holding the upside-down MG90S robotic arm from the top slats, with the servo driver board and glowing Arduino Nano breadboard mounted on the left wooden wall and VTM Studio Arm Control Mode running on the laptop in the background.
Custom Underside Swivel Mount (Upside-Down Arm Implementation): Moving the arm from an outside shoulder mount to this underside swivel mount hanging from the top slats takes the hanging weight of the arm off the first shoulder rotator at the arm's base—with the servo header board and Nano breadboard mounted on the left crate wall and VTM Studio Arm Control Mode running in the background.

Step 3: Add AI Vision (Using Your Laptop Cam or Phone with Iriun), Voice, and Servos

With your components mounted on your frame, you turn those physical computing projects into physical AI projects inside VTM Studio—without needing to buy expensive extra accessories:

  • 1. Use Your Laptop Camera or Add Your Cell Phone with Iriun in Vision Mode: You do not need to buy a dedicated USB camera to get started with AI vision. You can use the built-in webcam on your laptop, or you can add your cell phone as a camera using an app like Iriun Webcam (just as I have done on my own setup) and select it in the CAM: menu inside Vision Mode. Using your phone lets you position the camera anywhere on your wooden crate or frame while running HSV Lock color tracking, VTM Vision object detection, or Cloud Vision.
  • 2. Add Voice and Mouth Animation: Use your computer's microphone and speakers for click-to-talk faster-whisper voice input, local Piper ONNX spoken audio readback, and animated mouth feedback when talking with ARC and VAS-X.
  • 3. Power Servos from a Separate 5 V–6 V Supply in Arm Control Mode: When you connect a servo arm to Arm Control Mode, power the servo rail from an external 5.0 V–6.0 V DC power supply (sharing a common GND with your microcontroller) so moving multiple joints never drops your computer's USB voltage.
  • 4. Run Cloud or Local Edge AI Models in Model Forge Mode: Pair your Cloud AI keys with local models by using Model Forge Mode's 3-tab workflow to convert local Hugging Face .safetensors files to .gguf, quantize them to 4-bit Q4_K_M or 8-bit Q8_0, and load them into llama-server.

Frequently Asked Questions

Do I need to buy a dedicated USB camera to set up a physical AI lab in VTM Studio?

No. You can use the built-in webcam on your laptop, or you can use your cell phone as a camera with an app like Iriun Webcam and select it directly from the CAM: dropdown in VTM Studio's Vision Mode.

Does a physical AI lab build have to be a robot?

No. A Physical AI project can be any purposeful physical computing build—such as a sensor and camera station, a LoRa telemetry rig, or a voice-assisted desk monitor. And because P³ stands for Physical Prototyping Platform, your first builds can simply be for testing, learning, and trying things out—like figuring out where to mount a sensor or whether to mount an arm on the side of a wooden crate or under the top slats.

How does VTM Studio guide you from a circuit plan on screen to a physical P³ build?

In VTM Studio, you first plan your 2D breadboard layout in Layout Mode, check pinouts in Spec Mode, verify resistor math in Reference Mode, test voltage and continuity in Meter Mode, and flash C++ firmware in Code Mode to watch live gauges in Science Mode. Once the circuit works on the breadboard, you mount those components onto your P³ frame—where going from a plan on paper or screen to the actual physical build helps you work out real-world logistics.