ABX00087 UNO R4 WiFi Development Board
Fa'ailoga Kirikiti Fa'aaoga Arduino UNO R4 WiFi + ADXL345 + Edge
Uunaiga
O lenei pepa o loʻo tuʻuina atu ai le faʻaogaina atoatoa o galuega mo le fausiaina o se faiga e iloa ai fana kirikiti e faʻaaoga ai le Arduino UNO R4 WiFi ma le ADXL345 accelerometer ma Edge Impulse Studio. O le poloketi e aofia ai le aoina o faʻamatalaga faʻavavevave, aʻoaʻoina se faʻataʻitaʻiga aʻoaʻoga masini, ma faʻapipiʻi le faʻataʻitaʻiga aʻoaʻoina i tua i le Arduino mo le faʻavasegaina o taimi moni.
Ata kirikiti na iloiloina i lenei poloketi:
– Ufiufi taavale
– Ta'avale Sa'o
– Toso Pu
Laasaga 1: Meafaigaluega Manaomia
- Arduino UNO R4 WiFi
– ADXL345 Accelerometer (I2C)
– Uaea osooso
– Laupapa falaoa (filifiliga)
– USB Type-C uaea
Laasaga 2: Polokalama Manaoga
- Arduino IDE (fou)
- Edge Impulse Studio tala (fuafua)
- Edge Impulse CLI meafaigaluega (Node.js manaʻomia)
- Adafruit ADXL345 faletusi
Laasaga 3: Uea le ADXL345
Fa'afeso'ota'i le masini ADXL345 i le Arduino UNO R4 WiFi fa'apea:
VCC → 3.3V
GND → GND
SDA → SDA (A4)
SCL → SCL (A5)
CS → 3.3V (filifiliga, mo le I2C mode)
SDO → opeopea po o le GND
Laasaga 4: Saunia IDE Sensor
E faʻafefea ona faʻapipiʻi le Sensor Library i Arduino IDE?
Tatala Arduino IDE
Tatala Meafaigaluega → Puleaina Faletusi… ma fa'apipi'i: Adafruit ADXL345 Unified Adafruit Unified Sensor
(Afai e iai sau LSM6DSO poʻo MPU6050 nai lo: faʻapipiʻi SparkFun LSM6DSO , Adafruit LSM6DS poʻo MPU6050 e tusa ai.)
Laasaga 5: Arduino Sketch mo le aoina o faʻamaumauga
Tu'u le ata lenei i lau Arduino UNO R4 WiFi. O lo'o fa'asolo atu fa'amaumauga fa'avavevave ile CSV format (x,y,z) ile ~ 18 Hz mo Edge Impulse.
#fa'aofia
#fa'aofia
Adafruit_ADXL345_Unified accel =
Adafruit_ADXL345_Unified(12345);
fa'aletonu le seti() {
Serial.begin(115200);
afai (!accel.begin()) {
Serial.println(“Leai ADXL345 na maua”);
ae (1);
}
accel.setRange(ADXL345_RANGE_4_G);
}
avanoa gaogao() {
sensors_mea_t e;
accel.getEvent(&e);
Serial.print (e.acceleration.x);
Serial.print(“,”);
Serial.print(e.acceleration.y);
Serial.print(“,”);
Serial.println(e.acceleration.z);tuai(55); // ~18 Hz
}
Seti Edge Impulse
Laasaga 6: Fesoʻotaʻi ile Edge Impulse
- Tapuni Arduino Serial Monitor.
- Fa'atonu le fa'atonuga: edge-impulse-data-forwarder -frequency 18
- Tu'u igoa axis: accX, accY, accZ
- Ta'u lau masini: Arduino-Cricket-Board
- Faʻamaonia le fesoʻotaʻiga i Edge Impulse Studio i lalo o 'Meafaigaluega'.
Laasaga 7: Aoina o Fa'amaumauga
I Edge Impulse Studio → Mauaina faʻamatalaga:
– Meafaigaluega: Arduino-Cricket-Board
– Sensori: Accelerometer (3 axes)
– SampLe umi: 2000 ms (2 sekone)
– Faatelevave: 18 Hz
Faamaumau ia le itiiti ifo i le 40 samples i le vasega:
– Ufiufi taavale
– Ta'avale Sa'o
– Toso Pu
Aoina Faamatalaga Examples
Ufiufi taavale
Meafaigaluega: Arduino-Cricket-Board
Fa'ailoga: Ufiufi taavale
Sensor: Sensor e 3 tu'i (accX, accY, accZ)
Sample umi: 10000ms
Auala: 18 Hz
Example Raw Data:
accX -0.32
sa'o 9.61
accZ -0.12
Ta'avale Sa'o
Meafaigaluega: Arduino-Cricket-Board
Fa'ailoga: Ta'avale Sa'o
Sensor: Sensor e 3 tu'i (accX, accY, accZ)
Sample umi: 10000ms
Auala: 18 Hz
Example Raw Data:
accX 1.24
sa'o 8.93
accZ -0.42
Toso Fana
Meafaigaluega: Arduino-Cricket-Board
Fa'ailoga: Pull Shot
Sensor: Sensor e 3 tu'i (accX, accY, accZ)
Sample umi:10000 ms
Auala: 18 Hz
Example Raw Data:
accX 2.01
sa'o 7.84
accZ -0.63 
Laasaga 8: Impulse Design
Tatala Fausia lagona:
Poloka fa'aofi: Fa'amaumauga fa'asologa o taimi (3 axes).
Tele Fa'amalama: 1000 ms Fa'atuputeleina Fa'amalama (laa): 200 ms Fa'aaga: Axes, Magnitude (filifiliga), fa'avevesi 18.
Polokalama fa'agaioiga: Su'esu'ega fa'apitoa (a'o Fa'aaliga Fa'apitoa mo le lafo). Tele fa'amalama: 1000 ms Fa'atuputeleina fa'amalama (laa): 200 ms Fa'aaga: Axes, Magnitude (filifiliga), fa'atumauina fa'amuamua fa'aletonu uma.
Poloka a'oa'oga: Fa'avasegaina (Keras).
Kiliki Save impulse. 
Fa'atupu uiga:
Alu ile Spectral analysis, kiliki Save parameters, ona fa'atupu lea o foliga mo le seti a'oa'oga.
Aoao se tamaʻi faʻataʻitaʻiga
Alu i le Classifier (Keras) ma fa'aoga se fa'aoga fa'apitoa e pei o:
Feso'ota'iga neural: 1–2 papa mafiafia (eg, 60 → 30), ReLU
Vaitaimi: 40–60
Fuafuaga a'oa'oga: 0.001–0.005
Tele vaega: 32
Vaevae fa'amaumauga: 80/20 (to'ele / su'ega)
Faasaoina ma aʻoaʻo faʻamaumauga
Iloilo ma Siaki Fa'ata'ita'iga su'ega i le seti taofi.
Asiasia le matrix fenumiai; afai o lo'o felavasa'i le li'o ma luga, ao mai nisi fa'amatalaga 'ese'ese pe tweak
Fa'ailoga mata'utia (tele fa'amalama / fola pisapisao).
Laasaga 9: Faʻapipiʻi i Arduino
Alu i le Fa'atulagaina:
Filifili Arduino faletusi (C++ faletusi e aoga foi).
Fa'aagaoi le EON Compiler (pe a maua) e fa'aiti'itia le tele o fa'ata'ita'iga. 
Sii mai le .zip, ona i ai lea i le Arduino IDE: Sketch → Include Library → Add .ZIP Library… E faaopoopo le exampe pei o le Static buffer ma Fa'aauau i lalo File → Examples →
Lau Igoa Poloketi – Edge Impulse. Fa'ailoga tusi mo Arduino UNO EK R4 WiFi + ADXL345.
Laasaga 10: Arduino Inference Sketch
#fa'aofia
#fa'aofia
#fa'aofia // Sui i le ulutala Edge Impulse
Adafruit_ADXL345_Unified accel =
Adafruit_ADXL345_Unified(12345);
static bool debug_nn = sese;
fa'aletonu le seti() {
Serial.begin(115200);
a'o (!Serial) {}
afai (!accel.begin()) {
Serial.println(“SESE: ADXL345 le iloa”);
ae (1);
}
accel.setRange(ADXL345_RANGE_4_G);
}
avanoa gaogao() {
fa'afefe fa'afefe[EI_CLASSIFIER_DSP_INPUT_FRAME_SIZE] = {0};
mo (tele_t ix = 0; ix < EI_CLASSIFIER_DSP_INPUT_FRAME_SIZE; ix +=
3) {
uint64_t next_tick = micros() + (EI_CLASSIFIER_INTERVAL_MS *
1000);
sensors_mea_t e;
accel.getEvent(&e);
puipui[ix + 0] = e.vave.x;
pa'u[ix + 1] = e.vave.y;
pa'u[ix + 2] = e.acceleration.z;
int32_t faatali = (int32_t)(next_tick – micros());
afai (faatalitali > 0) tuaiMicroseconds(faatalitali);
}
faailo_t faailo;
int err = numpy::signal_from_buffer(buffer,
EI_CLASSIFIER_DSP_INPUT_FRAME_SIZE, &faailoga);
afai (sese != 0) toe fo'i;
ei_impulse_result_t result = {0};
EI_IMPULSE_ERROR res = run_classifier(&faailoga, &i'uga,
debug_nn);
afai (res != EI_IMPULSE_OK) toe foʻi;
mo (tele_t ix = 0; ix < EI_CLASSIFIER_LABEL_COUNT; ix++) {
ei_printf(“%s: %.3f “, taunuuga.faavasegaina[ix].igoa,
taunuuga.faavasegaina[ix].tau);
}
#afai EI_CLASSIFIER_AI_ANOMALY == 1
ei_printf(“anomaly: %.3f”, result.anomaly);
#endif
ei_printf(“\n”);
}
Galuega fa'auluampLe:
Fautuaga:
Taofi le EI_CLASSIFIER_INTERVAL_MS i le ogatasi ma lou fa'asolo atu o fa'amaumauga (fa'ata'ita'iga, 100 Hz → 10 ms). O le faletusi Edge Impulse e faʻapipiʻi otometi lenei mea mai lou manaʻo.
Afai e te mana'o e su'esu'e fa'aauau (fa'amalama fa'ase'e), amata mai le Fa'aauau exampe aofia ai ma le faletusi EI ma fesuiai i le ADXL345 faitau.
O le a matou faʻaopoopoina aʻoaʻoga vitio lata mai; seia oo i lena taimi, tumau pea - https://www.youtube.com/@RobuInlabs
Ma afai o loʻo i ai pea ni ou masalosaloga, e mafai ona e siakiina lenei vitio e Edged Impulse: https://www.youtube.com/watch?v=FseGCn-oBA0&t=468s
Pepa / Punaoa
 |
Arduino ABX00087 UNO R4 WiFi Development Board [pdf] Taiala mo Tagata Fa'aoga R4 WiFi, ADXL345, ABX00087 UNO R4 WiFi Atinaʻe Komiti, ABX00087, UNO R4 WiFi Atinaʻe Komiti, WiFi Atinaʻe Komiti, Komiti Atinaʻe, Komiti Faʻatonu |