Example Code for Firebeetle 2 ESP32-E - Get Pose Data

Hardware Preparation

• FireBeetle 2 ESP32-E (SKU: DFR0654) ×1
• Gravity: AI Vision Posture and Gesture Sensor(SKU: SEN0670) ×1
• PH2.0-4P Cable ×1
• USB Data Cable ×1

Software Preparation

• Download Arduino IDE: Click to Download Arduino IDE
• Install SDK: Visit the FireBeetle 2 ESP32-E WIKI page for SDK installation instructions
• Download Arduino Library: Click to download DFRobot_HumanPose and refer to the guide: How to Install a Library?

Wiring Diagram

I2C Wiring Diagram

Pin Connection Description:

• Sensor: + Pin --- (Connects to) --- Main Controller: 3V3
• Sensor: - Pin --- (Connects to) --- Main Controller: GND
• Sensor: SCL Pin --- (Connects to) --- Main Controller: 22/SCL
• Sensor: SDA Pin --- (Connects to) --- Main Controller: 21/SDA

UART Wiring Diagram (This connection method is used in the Sample Code)

Pin Connection Description:

• Sensor: + Pin --- (Connects to) --- Main Controller: 3V3
• Sensor: - Pin --- (Connects to) --- Main Controller: GND
• Sensor: RX Pin --- (Connects to) --- Main Controller: 26/D3
• Sensor: TX Pin --- (Connects to) --- Main Controller: 25/D2

When the connection method of the sensor is changed from I2C connection to UART connection, please check if the communication mode switch is also synchronously switched to UART.

Sample Code

Function: Initialize posture sensor via UART, set to pose mode, print ID, name, score, bounding box and 17 keypoint coordinates.

#include <DFRobot_HumanPose.h>
#define HUMANPOSE_COMM_UART  // Use UART communication
// #define HUMANPOSE_COMM_I2C    // Use I2C communication

#if defined(HUMANPOSE_COMM_UART)
/* ---------------------------------------------------------------------------------------------------------------------
 * Hardware connection table:
 *    Sensor Pin |        MCU Pin        | Leonardo/Mega2560/M0 |    UNO    | ESP8266 | ESP32 |  microbit  |   m0  |
 *     VCC       |        3.3V/5V         |        VCC           |    VCC    |   VCC   |  VCC  |     X      |  vcc  |
 *     GND       |         GND            |        GND           |    GND    |   GND   |  GND  |     X      |  gnd  |
 *     RX        |        MCU TX          |     Serial1 TX1      |     5     |   5/D6  |  26/D3|     X      |  tx1  |
 *     TX        |        MCU RX          |     Serial1 RX1      |     4     |   4/D7  |  25/D2|     X      |  rx1  |
 * ----------------------------------------------------------------------------------------------------------------------*/
// Initialize UART communication: Serial1, baud rate 9600, RX pin 25, TX pin 26 (ESP32)
#if defined(ARDUINO_AVR_UNO) || defined(ESP8266)
#include <SoftwareSerial.h>
SoftwareSerial         mySerial(4, 5);
DFRobot_HumanPose_UART humanPose(&mySerial, 921600);
#elif defined(ESP32)
DFRobot_HumanPose_UART humanPose(&Serial1, 921600, /*RX pin*/ 25, /*TX pin*/ 26);
#else
DFRobot_HumanPose_UART humanPose(&Serial1, 921600);
#endif
#elif defined(HUMANPOSE_COMM_I2C)

const uint8_t I2C_ADDR = 0x3A;
// Initialize I2C communication: Wire, I2C address 0x3A
DFRobot_HumanPose_I2C humanPose(&Wire, I2C_ADDR);
#else
#error "Please define HUMANPOSE_COMM_UART or HUMANPOSE_COMM_I2C"
#endif

void setup()
{
  // Initialize serial port for debug output
  Serial.begin(115200);

  // Initialize sensor, retry if failed
  while (!humanPose.begin()) {
    Serial.println("Sensor init fail!");
    delay(1000);
  }
  Serial.println("Sensor init success!");

  // Set detection model: eHand (hand detection) or ePose (human pose detection)
  humanPose.setModelType(DFRobot_HumanPose::ePose);
}

void loop()
{
  // Get detection results
  if (humanPose.getResult() == DFRobot_HumanPose::eOK) {
    Serial.println("getPoseResult success");
    while (humanPose.availableResult()) {
      PoseResult *result = static_cast<PoseResult *>(humanPose.popResult());
      Serial.println("id: " + String(result->id));
      Serial.println("name: " + result->name);
     
      Serial.println("score: " + String(result->score));
      Serial.println("xLeft: " + String(result->xLeft));
      Serial.println("yTop: " + String(result->yTop));
      Serial.println("width: " + String(result->width));
      Serial.println("height: " + String(result->height));
      Serial.println("nose: " + String(result->nose.x) + ", " + String(result->nose.y));
      Serial.println("leye: " + String(result->leye.x) + ", " + String(result->leye.y));
      Serial.println("reye: " + String(result->reye.x) + ", " + String(result->reye.y));
      Serial.println("lear: " + String(result->lear.x) + ", " + String(result->lear.y));
      Serial.println("rear: " + String(result->rear.x) + ", " + String(result->rear.y));
      Serial.println("lshoulder: " + String(result->lshoulder.x) + ", " + String(result->lshoulder.y));
      Serial.println("rshoulder: " + String(result->rshoulder.x) + ", " + String(result->rshoulder.y));
      Serial.println("lelbow: " + String(result->lelbow.x) + ", " + String(result->lelbow.y));
      Serial.println("relbow: " + String(result->relbow.x) + ", " + String(result->relbow.y));
      Serial.println("lwrist: " + String(result->lwrist.x) + ", " + String(result->lwrist.y));
      Serial.println("rwrist: " + String(result->rwrist.x) + ", " + String(result->rwrist.y));
      Serial.println("lhip: " + String(result->lhip.x) + ", " + String(result->lhip.y));
      Serial.println("rhip: " + String(result->rhip.x) + ", " + String(result->rhip.y));
      Serial.println("lknee: " + String(result->lknee.x) + ", " + String(result->lknee.y));
      Serial.println("rknee: " + String(result->rknee.x) + ", " + String(result->rknee.y));
      Serial.println("lankle: " + String(result->lankle.x) + ", " + String(result->lankle.y));
      Serial.println("rankle: " + String(result->rankle.x) + ", " + String(result->rankle.y));
      Serial.println("--------------------------------");
    }
  } else {
    Serial.println("getResult fail");
  }

  // Delay to avoid output too fast
  delay(100);
}

Result

After successful initialization, the sensor cyclically detects human poses, printing detection prompt, ID, name, confidence, bounding box and 17 keypoint coordinates in real time; prints failure prompt if detection fails, updating every 100ms.

As shown in the serial port output information in the following figure, an unknown posture has been recognized.

As shown in the serial port output information below, the learned pose is recognized as' stand'.