martedì 30 marzo 2021
Lettore file UBX in Python
uno script scritto di fretta per risolvere un problema di debug sulla trasmissione via radio di un file ubx
magari puo' servire a qualcun altro
=================================================
def crc_check(valore):
global ck_a
global ck_b
ck_a = (ck_a+valore) % 256
ck_b = (ck_b+ck_a) % 256
#print "CK_A : " + str(hex(ck_a)) + " CK_B : " + str(hex(ck_b))
#file = open('acm0_remoto.ubx', 'r')
#file = open('usb0.ubx', 'r')
file = open('arcetri_ucenter.ubx', 'r')
#file = open('orto.ubx', 'r')
off=0
fine=0
ck_a = 0
ck_b = 0
good = 0
bad = 0
pacchetto = -1
while 1:
char = file.read(1)
off=off+1
pacchetto = ""
if ord(char) == 181: #B5
pacchetto = pacchetto + char
if (off == (fine+1)):
print "OK"
inizio = off -1
ck_a = 0
ck_b = 0
char = file.read(1)
off=off+1
#pacchetto = pacchetto + char
if ord(char) == 98: #62
#
if (inizio == fine + 1):
print "////////////////////// OK ////////////////"
print "/////////////////// " + hex(lungh) +" ////////////////"
#good = good + 1
else:
print "******************** ERRORE **************"
print "*****************DIFF " + hex(inizio-fine) + "*********"
#somma_diff = somma_diff + (inizio-fine)
#bad = bad+1
print "================================================="
print "======GOOD :" + str(good) + "==========BAD :"+ str(bad) + "==========================="
print "================================================="
pacchetto = pacchetto + char
print "Offset inizio: " + hex(inizio)
char = file.read(1)
pacchetto = pacchetto + char
off=off+1
print "Message Class: " + hex(ord(char)),
if (ord(char) == 1):
print " NAV ",
if (ord(char) == 2):
print " RXM ",
if (ord(char) == 4):
print " INF ",
if (ord(char) == 5):
print " ACK ",
if (ord(char) == 6):
print " CFG ",
if (ord(char) == 10):
print " MON ",
if (ord(char) == 11):
print " AID ",
if (ord(char) == 13):
print " TIM ",
classe = ord(char)
crc_check(ord(char))
char = file.read(1)
pacchetto = pacchetto + char
off=off+1
print " Message ID: " + hex(ord(char)),
crc_check(ord(char))
if (classe == 2):
if (ord(char)) == 0x10:
print " RAW "
if (ord(char)) == 0x11:
print " SFRB "
if (ord(char)) == 0x12:
print " SFRBX "
if (ord(char)) == 0x13:
print " RAWX "
char = file.read(1)
pacchetto = pacchetto + char
off=off+1
print "Payload Length High: " + hex(ord(char))# + " " + str(ord(char))
lungh = ord(char)
crc_check(ord(char))
char = file.read(1)
pacchetto = pacchetto + char
off=off+1
print "Payload Length Low: " + hex(ord(char))# + " " + str(ord(char))
lungh = (256*ord(char)) + lungh
print "Lunghezza :" + str(hex(lungh))#+ " " + str(lungh)
crc_check(ord(char))
for t in range(lungh):
char = file.read(1)
print hex(ord(char)),
pacchetto = pacchetto + char
off=off+1
crc_check(ord(char))
char = file.read(1)
pacchetto = pacchetto + char
off=off+1
crc_1 = ord(char)
char = file.read(1)
pacchetto = pacchetto + char
off=off+1
print "CRC 1: " + hex(crc_1) +" CRC 2: " + hex(ord(char))
crc_2 = ord(char)
print "CK_A :" + hex(ck_a) + " CK_B : " + hex(ck_b)
if ((ck_a == crc_1) and (ck_b == crc_2)):
print "CRC OK"
good = good + 1
corretto = open('corretto.ubx', 'a')
corretto.write(pacchetto)
corretto.close()
else:
print "Errore CRC"
bad = bad + 1
fine = off-1
print "Fine :" + hex(fine)
try:
input()
except SyntaxError:
pass
giovedì 18 febbraio 2021
Motorizzare telescopio Auriga Skywatcher
Per motorizzare un asse di una montatura EQ1 di un rifrattore 70/900 della Skywatcher ho provato a comprare su Aliexpress un motore delle Celestron (non ho idea se sia un Celestron originale e non avevo idea se poteva essere montato da una equatoriale Skywatcher)
Per prima cosa si smonta la vite sull'asse della latitudine
martedì 16 febbraio 2021
Populele 2
Ho avuto in prestito un Populele 2, ukulele "smart" della Xiaomi
Non ho mai suonato ukulele e quindi non ho esperienza precedente. Di sicuro e' fatto di plastica e non di carbonio come viene indicato nella pubblicita'. In merito al suono non saprei dire ma sicuramente la tastiera di un ukulele e' microscopica
Per quanto riguarda la parte smart e' stata una mezza delusione. Di fatto l'ukulele ha solo dei led affogati nel manico (72 per essere precisi) che si illuminano per indicare dove mettere le dita collegati ad un ricevitore bluetooth. Tutta la parte di riconoscimento delle note e degli accordi e' tutta demandata alla app che risiede sul telefono (e che ovviamente non si attiva se non trova il pairing in modo da non poter usata con altri ukulele non smart)
a questo link https://github.com/Furikuda/Hackulele c'e' un interessante hack che permette di accendere a piacere i led ed una analisi degli integrati presenti (risultano esserci un microcontrollore Dialog 14580 ed un controllore per la matrice di led
Per finire la cosa curiosa e' che la ditta consiglia, a seguito di numerose critiche, di non scaricare la propria app dal Google Store ma di installare l'apk dal proprio website...comportamento discutibile
giovedì 11 febbraio 2021
Emulatore Android su Apple M1
martedì 2 febbraio 2021
Anomaly detection con Tensorflow Lite e Arduino Nano BLE 33
Quest post riguarda un esperimento per utilizzare una rete neurale su microcontrollori
Nello specifico e' stato provato una Anomaly Detection perche' spesso i processi geologici sono difficilmente ripetibili (e se si ripetono non sono nelle stesse condizioni..basti vedere le frane in cui si possono avere neoformazioni) quindi e' praticamente impossibile definire un set di addestramento per la rete neurale....la cosa piu' conveniente e' vedere se ci sono anomalie all'interno di uno stream di dati da un sensore.
La rete neurale per la Anomaly Detectione e' stata impostata mediante un Autoencoder partendo da questo tutorial https://www.tensorflow.org/tutorials/generative/autoencoder
Per i dati e' stata scelta una Arduino Nano 33 BLE che ha a bordo una accelerometro ed un giroscopio triassiale piu' un sensore di temperatura (dato che serve a gestire la deriva termica dei sensori MEMS)
Riguardo alla scheda un paio di note
1) il sensore di temperatura e' montato sullo stesso PCB del microcontrollore e quindi risente del riscaldamento, seppur minimo, del microcontrollore. La cosa piu' conveniente e' iniziare la registrazione quanto tutto il sistema e' andato in equilibrio termico
2) la velocita' di acquisizione del sensore termico e' molto piu' lenta dell'accelerometro. Se si interroga il sensore termico troppo velocemente il sistema si blocca (e non dice nemmeno il motivo del blocco..)
Lo sketch di acquisizione dati e' banale e semplicemente manda sulla seriale delle stringhe formattate in modo da costituire un semplice file csv nel formato ax,ay,az,gx,gy,gz,temp
===================================================
nano-33-sense-serial-example.ino
Copyright (c) 2020 Dale Giancono. All rights reserved..
This program outputs all raw sensor data from the Arduino Nano 33 BLE
Sense board via serial at a 20Hz rate. It also calculates the RMS
value of the microphone buffer and outputs that data. It is intended
as a quick way to become familiar with some of the sensor libraries
available with the Nano 33 BLE Sense, and highlight some of the
difficulties when using a super loop architecture with this board.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <https://www.gnu.org/licenses/>.
*/
/**********/
/*INCLUDES*/
/**********/
/* For MP34DT05 microphone */
#include <PDM.h>
/* For LSM9DS1 9-axis IMU sensor */
#include <Arduino_LSM9DS1.h>
/* For APDS9960 Gesture, light, and proximity sensor */
#include <Arduino_APDS9960.h>
/* For LPS22HB barometric barometricPressure sensor */
#include <Arduino_LPS22HB.h>
/* For HTS221 Temperature and humidity sensor */
#include <Arduino_HTS221.h>
#include <arm_math.h>
/********/
/*MACROS*/
/********/
/* Set these macros to true with you want the output plotted in a way that
* can be viewed with serial plotter. Having them all true creates a pretty
* meaningless graph, as the scaling will be way off for each sensor, and there
* will be too much data to view */
#define SERIAL_PLOT_MP34DT05 (false)
#define SERIAL_PLOT_LSM9DS1 (true)
#define SERIAL_PLOT_APDS9960 (false)
#define SERIAL_PLOT_LPS22HB (false)
#define SERIAL_PLOT_HTS221 (true)
/* This value was also used in the PDM example, seems like a good enough reason to
* continue using it. With this value and 16kHz sampling frequency, the RMS sampling
* period will be 16mS */
#define MICROPHONE_BUFFER_SIZE_IN_WORDS (256U)
#define MICROPHONE_BUFFER_SIZE_IN_BYTES (MICROPHONE_BUFFER_SIZE_IN_WORDS * sizeof(int16_t))
/******************/
/*LOCAL VARIABLES*/
/******************/
/******************/
/*GLOBAL VARIABLES*/
/******************/
/* MP34DT05 Microphone data buffer with a bit depth of 16. Also a variable for the RMS value */
int16_t microphoneBuffer[MICROPHONE_BUFFER_SIZE_IN_WORDS];
int16_t microphoneRMSValue;
/* variables to hold LSM9DS1 accelerometer data */
float accelerometerX, accelerometerY, accelerometerZ;
/* variables to hold LSM9DS1 gyroscope data */
float gyroscopeX, gyroscopeY, gyroscopeZ;
/* variables to hold LSM9DS1 magnetic data */
float magneticX, magneticY, magneticZ;
/* variables to hold LPS22HB barometric pressure data */
float barometricPressure;
/* variables to hold APDS9960 proximity, gesture and colour data */
int proximity, gesture, colourR, colourG, colourB;
/* variables to hold HTS221 temperature and humidity data */
float temperature, humidity;
/* Used to count 1000ms intervals in loop() */
int oldMillis;
int newMillis;
/* Used as a simple flag to know when microphone buffer is full and RMS value
* can be computed. */
bool microphoneBufferReadyFlag;
/****************************/
/*LOCAL FUNCTION PROTOTYPES*/
/****************************/
/****************************/
/*GLOBAL FUNCTION PROTOTYPES*/
/****************************/
/* This function is called each time PDM data is available. It will be used to fill the
* microphone buffer that we will then use to calculate RMS values */
void Microphone_availablePDMDataCallback(void);
/* This function computes the RMS value based on the data contained within the microphoneBuffer
* If the microphone buffer has a word length of 256, and the sample rate for the microphone is 16kHz,
* then this RMS value is taken over (1/16000)*256 = 16mS */
void Micophone_computeRMSValue(void);
/****************************/
/*IMPLEMENTATION*/
/****************************/
void setup()
{
/* Serial setup for UART debugging */
Serial.begin(115200);
/* Wait for serial to be available */
while(!Serial);
/* PDM setup for MP34DT05 microphone */
/* configure the data receive callback to transfer data to local buffer */
PDM.onReceive(Microphone_availablePDMDataCallback);
/* Initialise single PDM channel with a 16KHz sample rate (only 16kHz or 44.1kHz available */
if (!PDM.begin(1, 16000))
{
Serial.println("Failed to start PDM!");
/* Hacky way of stopping program executation in event of failure. */
while(1);
}
else
{
/* Gain values can be from 0 to 80 (around 38db). Check out nrf_pdm.h
* from the nRF528x-mbedos core to confirm this. */
/* This has to be done after PDM.begin() is called as begin() always
* sets the gain as the default PDM.h value (20).
*/
PDM.setGain(50);
}
/* IMU setup for LSM9DS1*/
/* default setup has all sensors active in continous mode. Sample rates
* are as follows: magneticFieldSampleRate = 20Hz, gyroscopeYroscopeSampleRate = 109Hz,
* accelerationSampleRate = 109Hz */
if (!IMU.begin())
{
Serial.println("Failed to initialize IMU!");
/* Hacky way of stopping program executation in event of failure. */
while(1);
}
/* Set sensitivity from 0 to 100. Higher is more sensitive. In
* my experience it requires quite a bit of experimentation to
* get this right, as if it is too sensitive gestures will always
* register as GESTURE_DOWN or GESTURE_UP and never GESTURE_LEFT or
* GESTURE_RIGHT. This can be called before APDS.begin() as it just
* sets an internal sensitivity value.*/
APDS.setGestureSensitivity(50);
if (!APDS.begin())
{
Serial.println("Error initializing APDS9960 sensor.");
/* Hacky way of stopping program executation in event of failure. */
while(1);
}
/* As per Arduino_APDS9960.h, 0=100%, 1=150%, 2=200%, 3=300%. Obviously more
* boost results in more power consumption. */
APDS.setLEDBoost(0);
/* Barometric sensor setup for LPS22HB*/
if (!BARO.begin())
{
Serial.println("Failed to initialize barometricPressure sensor!");
while (1);
}
/* Temperature/Humidity sensor setup for HTS221*/
if (!HTS.begin())
{
Serial.println("Failed to initialize humidity temperature sensor!");
/* Hacky way of stopping program executation in event of failure. */
while(1);
}
/* Initialise timing variables. */
oldMillis = 0;
newMillis = 0;
/* Initialise micophone buffer ready flag */
microphoneBufferReadyFlag = false;
}
void loop()
{
/* The sensors that use I2C must be checked to see if data is available, so
* this is checked each loop. This include the IMU and Gesture/light/proximity
* sensors. Other sensors (barometric pressure and temperature/humidity)
* will give a value when we ask for it. These values are requested each
* 1000ms using millis() in a hacky way, but it works.
*
* Data is output via serial every 50ms (20Hz). There is no good way to plot of
* the data from the sensors together due the differing sample rates, but this
* represented a decent compromise as changes will still be observable in all
* sensor data.
*/
/* Get the new millis() value which helps time serial plotting and getting
* of pressure, temperature, and humidity values. */
newMillis = millis();
/* Every 50ms plot all data to serial plotter. */
if((newMillis - oldMillis) % 50)
{
#if (SERIAL_PLOT_MP34DT05 == true)
Serial.printf("%d,", microphoneRMSValue);
#endif
#if (SERIAL_PLOT_LSM9DS1 == true)
Serial.printf("%f,%f,%f,", accelerometerX, accelerometerY, accelerometerZ);
Serial.printf("%f,%f,%f,", gyroscopeX, gyroscopeY, gyroscopeZ);
//Serial.printf("%f,%f,%f,", magneticX, magneticY, magneticZ);
#endif
#if (SERIAL_PLOT_LPS22HB == true)
Serial.printf("%f,", barometricPressure);
#endif
#if (SERIAL_PLOT_APDS9960 == true)
Serial.printf("%d,%d,%d,%d,%d,", proximity, gesture, colourR, colourG, colourB);
#endif
#if (SERIAL_PLOT_HTS221 == true)
//Serial.printf("%f, %f", temperature, humidity);
Serial.printf("%f", temperature);
#endif
Serial.println();
}
/* Every 1000ms get the pressure, temperature, and humidity data */
if((newMillis - oldMillis) > 1000)
{
barometricPressure = BARO.readPressure();
temperature = HTS.readTemperature();
humidity = HTS.readHumidity();
oldMillis = newMillis;
}
/* If new acceleration data is available on the LSM9DS1 get the data.*/
if(IMU.accelerationAvailable())
{
IMU.readAcceleration(accelerometerX, accelerometerY, accelerometerZ);
}
/* If new gyroscope data is available on the LSM9DS1 get the data.*/
if(IMU.gyroscopeAvailable())
{
IMU.readGyroscope(gyroscopeX, gyroscopeY, gyroscopeZ);
}
/* If new magnetic data is available on the LSM9DS1 get the data.*/
if (IMU.magneticFieldAvailable())
{
IMU.readMagneticField(magneticX, magneticY, magneticZ);
}
/* If new proximity data is available on the APDS9960 get the data.*/
if (APDS.proximityAvailable())
{
proximity = APDS.readProximity();
}
/* If new colour data is available on the APDS9960 get the data.*/
if (APDS.colorAvailable())
{
APDS.readColor(colourR, colourG, colourB);
}
/* If new gesture data is available on the APDS9960 get the data.*/
if (APDS.gestureAvailable())
{
gesture = APDS.readGesture();
}
/* If the microphone buffer is full, compute the RMS value */
if(microphoneBufferReadyFlag)
{
Micophone_computeRMSValue();
microphoneBufferReadyFlag = false;
}
}
void Microphone_availablePDMDataCallback()
{
// query the number of bytes available
int bytesAvailable = PDM.available();
if(bytesAvailable == MICROPHONE_BUFFER_SIZE_IN_BYTES)
{
PDM.read(microphoneBuffer, bytesAvailable);
microphoneBufferReadyFlag = true;
}
}
void Micophone_computeRMSValue(void)
{
//arm_rms_q15((q15_t*)microphoneBuffer, MICROPHONE_BUFFER_SIZE_IN_WORDS, (q15_t*)µphoneRMSValue);
}
===================================================
i dati sono stati registrati su due file in cui uno e' il train delle rete neurale (sistema completamente statico) mentre sul secondo e' stato registrato un evento con un colpetto sull'asse Z
le elaborazioni sono state effettuate mediante il seguente Notebook di Colab
Train 26000 dati |
Evento |
i dati sono stati normalizzati e sono stati inseriti nella seguente neurale
Threshold: 0.003555521
questo e' l'evento individuato dalla rete neurale nel dataset di test
la rete e' stata convertita nel formato di Tensorflow Lite e quindi tramite il comando xxd e' stata convertita in un array di byte per l'utilizzo in Tensorflow per microcontrollori
Lo sketch di Arduino e' stato ripreso dagli esempi del libro
TinyML: Machine Learning With Tensorflow Lite on Arduino and Ultra-Low-Power Microcontrollers: Machine Learning with TensorFlow on Arduino, and Ultra-Low Power Micro-Controllers
di cui si trovano gli esempi su GitHub
Si copia il risultato dell'array derivante da xxd sul file model.h
Lo sketch che ho utilizzato per invocare la rete neurale e' il seguente
=====================================================
IMU Classifier
This example uses the on-board IMU to start reading acceleration and gyroscope
data from on-board IMU, once enough samples are read, it then uses a
TensorFlow Lite (Micro) model to try to classify the movement as a known gesture.
Note: The direct use of C/C++ pointers, namespaces, and dynamic memory is generally
discouraged in Arduino examples, and in the future the TensorFlowLite library
might change to make the sketch simpler.
The circuit:
- Arduino Nano 33 BLE or Arduino Nano 33 BLE Sense board.
Created by Don Coleman, Sandeep Mistry
Modified by Dominic Pajak, Sandeep Mistry
This example code is in the public domain.
*/
#include <Arduino_LSM9DS1.h>
#include <Arduino_HTS221.h>
#include <TensorFlowLite.h>
#include <tensorflow/lite/micro/all_ops_resolver.h>
#include <tensorflow/lite/micro/micro_error_reporter.h>
#include <tensorflow/lite/micro/micro_interpreter.h>
#include <tensorflow/lite/schema/schema_generated.h>
#include <tensorflow/lite/version.h>
#include "model.h"
tflite::MicroErrorReporter tflErrorReporter;
tflite::AllOpsResolver tflOpsResolver;
const tflite::Model* tflModel = nullptr;
tflite::MicroInterpreter* tflInterpreter = nullptr;
TfLiteTensor* tflInputTensor = nullptr;
TfLiteTensor* tflOutputTensor = nullptr;
constexpr int tensorArenaSize = 120 * 1024;
byte tensorArena[tensorArenaSize];
int oldMillis;
int newMillis;
float oldvalue;
void setup() {
oldMillis = 0;
newMillis = 0;
oldvalue = 0.0;
Serial.begin(9600);
while (!Serial);
if (!IMU.begin()) {
Serial.println("Failed to initialize IMU!");
while (1);
}
if (!HTS.begin())
{
Serial.println("Failed to initialize humidity temperature sensor!");
while(1);
}
tflModel = tflite::GetModel(drive_My_Drive_autoencode_modello_model_tflite);
if (tflModel->version() != TFLITE_SCHEMA_VERSION) {
Serial.println("Model schema mismatch!");
while (1);
}
// Create an interpreter to run the model
tflInterpreter = new tflite::MicroInterpreter(tflModel, tflOpsResolver, tensorArena, tensorArenaSize, &tflErrorReporter);
// Allocate memory for the model's input and output tensors
tflInterpreter->AllocateTensors();
// Get pointers for the model's input and output tensors
tflInputTensor = tflInterpreter->input(0);
tflOutputTensor = tflInterpreter->output(0);
}
void loop() {
float aX, aY, aZ, gX, gY, gZ,temperature;
newMillis = millis();
if((newMillis - oldMillis) > 1000)
{
temperature = HTS.readTemperature();
oldMillis = newMillis;
}
if (IMU.accelerationAvailable() && IMU.gyroscopeAvailable()) {
IMU.readAcceleration(aX, aY, aZ);
IMU.readGyroscope(gX, gY, gZ);
tflInputTensor->data.f[0] = (aX + 4.0) / 8.0;
tflInputTensor->data.f[1] = (aY + 4.0) / 8.0;
tflInputTensor->data.f[2] = (aZ + 4.0) / 8.0;
tflInputTensor->data.f[3] = (gX + 2000.0) / 4000.0;
tflInputTensor->data.f[4] = (gY + 2000.0) / 4000.0;
tflInputTensor->data.f[5] = (gZ + 2000.0) / 4000.0;
tflInputTensor->data.f[6] = (temperature) / 50.0;
TfLiteStatus invokeStatus = tflInterpreter->Invoke();
if (invokeStatus != kTfLiteOk) {
Serial.println("Invoke failed!");
while (1);
return;
}
//Serial.println(tflOutputTensor->data.f[0], 6);
if (tflOutputTensor->data.f[0] > oldvalue+0.00355){
Serial.println("1");
}
else{
Serial.println("0");
}
oldvalue = tflOutputTensor->data.f[0];
}
}
=====================================================
Debugger integrato ESP32S3
Aggiornamento In realta' il Jtag USB funziona anche sui moduli cinesi Il problema risiede nell'ID USB della porta Jtag. Nel modulo...
-
In questo post viene indicato come creare uno scatterplot dinamico basato da dati ripresi da un file csv (nel dettaglio il file csv e' c...
-
Questo post e' a seguito di quanto gia' visto nella precedente prova Lo scopo e' sempre il solito: creare un sistema che permet...
-
La scheda ESP32-2432S028R monta un Esp Dev Module con uno schermo TFT a driver ILI9341 di 320x240 pixels 16 bit colore.Il sito di riferiment...