Example Code for Arduino - Custom Gesture Recognition

Hardware Preparation

• Gravity: AI Posture and Gesture Sensor (SKU: SEN0670) ×1
• FireBeetle 2 ESP32‑E (SKU: DFR0654)
• USB Type‑C cable ×1
• PC ×1

Software Preparation

• Programming environment: Click to download Arduino IDE
• Download and install the sensor library: DFRobot_HumanPose
• Please click here to view: How to install a library?

Wiring Diagram

I²C Connection (The following examples all use I²C communication)

Connection Description:

UART Connection

Connection Description:

Note:

• The default communication method of the sensor is I²C. If you need to switch to UART, use the on‑board DIP switch to change the communication method.
• After switching, the sensor must be powered off and on again for the change to take effect.

Sample Code

Before starting this example, make sure the sensor has already learned custom gestures using the host computer software. If no gestures have been learned beforehand, the sensor will recognise them as unknown gestures.

Function: Drive gesture sensor via I2C, output gesture type, confidence and position data.

#include <DFRobot_HumanPose.h>

// I2C communication by default
const uint8_t I2C_ADDR = 0x3A;
DFRobot_HumanPose_I2C humanPose(&Wire, I2C_ADDR);

void setup()
{
  Serial.begin(115200);

  // 1. Initialise the sensor
  while (!humanPose.begin()) {
    Serial.println("Sensor initialisation failed, please check the wiring!");
    delay(1000);
  }
  Serial.println("Sensor initialisation successful!");

  // 2. Set to "hand detection" mode
  humanPose.setModelType(DFRobot_HumanPose::eHand);
}

void loop()
{
  // 3. Get the detection result
  if (humanPose.getResult() == DFRobot_HumanPose::eOK) {
    // Iterate through all detected hands/gestures
    while (humanPose.availableResult()) {
      Result *result = humanPose.popResult();
      if (result) {
        // Logic: id=0 means not learned, id≠0 means learned
        if (result->id == 0) {
          Serial.println("[Unknown gesture]");
        } else {
          Serial.print("[Learned gesture] Name: ");
          Serial.println(result->name); // Print the name of the learned gesture
        }
        
        // Print common information: confidence, position
        Serial.print("Confidence: "); Serial.println(result->score);
        Serial.print("Position: left="); Serial.print(result->xLeft);
        Serial.print(" top="); Serial.print(result->yTop);
        Serial.print(" width="); Serial.print(result->width);
        Serial.print(" height="); Serial.println(result->height);
        Serial.println("------------------------");
      }
    }
  }

  delay(100); // 100ms delay to avoid spamming the serial monitor
}

Result:

The sensor automatically recognises gestures and can distinguish between learned and unlearned gestures.