OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz...

29
OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060 Juan Fernando Huertas B. - 1124820

Transcript of OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz...

Page 1: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

OBJECT RECOGNITION USING ULTRASONIC

SENSOR

Jeniffer Arias Sarasty – 1110099Pedro Luis Muñoz Murillo – 1210061

Cristian David Munera Cardona – 1210060Juan Fernando Huertas B. - 1124820

Page 2: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

The digital signal processing is an area that has been making progress in recent years, thanks to the development of the devices used for this purpose, such as integrated circuits or computers.

This work is focused on the proposal for implementation in hardware and software of a recognize objects by ultrasound. The hardware is composed by the step of sensing and signal conditioning, then by analyzing and signals processing the object will be classified.

INTRODUCTION

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 3: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Real aplications of ultrasound

• Ultrasonic obstacle detector for blind people• Object detection radars.• Measuring distances• Internal characterization of materials• Diagnostic ultrasound imaging in general hospital• detectors cracks in metallic materials

Page 4: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

GENERAL

Design a hardware and software system recognition of volumetric objects by their shape

OBJECTIVES

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 5: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

SPECIFIC

• Select the appropriate sensor that meets the project requirements.

• Study and implement the proper way to the conditioning and representation of the signals from the sensor.

• Implement a communication protocol between the signal acquisition system and the computer.

OBJECTIVES

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 6: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

SPECIFIC

• Develop a script with the proper technique for analyzing the information obtained from the data acquisition.

• Designing the GUI that shows the shape of the sensing object.

OBJECTIVES

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 7: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

The solution is presented in 5 stages

METHODOLOGY

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Graphical User Interface (GUI)

Processign

Data Collection

Signal Conditioning

Sensing

Page 8: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

They can point to two types of gauges, which have a transmitter and a separate receiver, and the alternating function, through a switching circuit, on the same transmitter / receiver piezo.

Sensing

Page 9: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

A scanning technique in which the object via a sensor wave which is reflected on the object is output is used, which causes an echo signal to be generated changes in characteristics.

For the transmitter will apply a pulse train of 40 KHz, for it to emit a signal which is then reflected in the object to be identified, generating an echo is detected by the receiver

Sensing

Page 10: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Change the surface characteristics as reflected.

Ultrasound propagation

Page 11: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

For data acquisition the Arduino Mega 2560, which is in charge of generating 9 pulses 40 kHz is used for the transmitter. Then through an analogue channel signal from the receiver is taken, which is converted to digital format (10 bits) for subsequent communication with Matlab through serial port.

Data Collection

Page 12: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Arduino-Matlab communication

Page 13: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Signal Conditioning

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Noise Removal Amplifying the signal

obtained Information extraction

Problems:

Page 14: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

High Pass Filter

Signal Conditioning

𝑓 𝑐=1

2𝜋 𝑅𝐶=1.3𝐾𝐻𝑧

Envelope Detector Section

Voltage proportional to retrieve a message

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 15: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Signal Conditioning

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 16: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Signal processingdigital filter

it was implemented a digital filter to eliminate the ripple in the obtained signal

Page 17: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Obtained signals for a plane Obtained signals for a corner

Obtained signals for a cilinder

Signals obtained

Page 18: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Characteristics extraction

in several works consulted is used temporal analysis of signals, getting nice features for further classification

Page 19: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Characteristics extraction

Page 20: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Artificial Neural Network

MLP type network is widely used for classification in different applications. There are several algorithms for training this network

• backpropagation algorithm• gradient descent algorithm with variable

alpha• Levenberg Marquardt algorithm

Page 21: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Artificial Neural Network

Page 22: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Artificial Neural Network

Page 23: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Graphical User Interface

Page 24: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Test protocol

Page 25: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Detection of three different shaped objects (plane, cylinder, corner) with inexpensive devices.

Object detection sizes.

It can be used as a basic prototype for a navigation system.

When used as a neural network processing, you can train better this network by adding more sensors so that a better description of the object is obtained

Scope

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 26: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Away for good signal acquisition was established by 30 cm. Less distance for interference due to echo presented.

Location of the object relative to the sensor: For a correct reading of the sensor, the object must be located so as to produce a signal reflection as only it has a sensor

When acquiring the object data must be in a space where there are no other objects that may cause echoes affecting the acquisition.

Shape between different objects must be very marked for the received echo is different, this is because the use of a single sensor.

Limitations

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 27: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Analysis and processing of ultrasonic signals to classify surfaces:

The recognition is done through an ultrasonic sensor rotation. The areas are characterized edges, corners, walls and empty.

Signal characterization Wavelet Transform.

performance compared to other solutions

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 28: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

For good operation of the neural network must acquire signals so that they can obtain certain characteristics, it is essential for a good filtering the signal from the sensor.

The combination of analog filters with digital filters is a way to eliminate unwanted signals such as ripple receiver signal acquired.

Implementing digital filters reduces costs and facilitates the signal processing as this is done by software, not by hardware.

Conclusions

OBJECT RECOGNITION USING ULTRASONIC SENSOR

Page 29: OBJECT RECOGNITION USING ULTRASONIC SENSOR Jeniffer Arias Sarasty – 1110099 Pedro Luis Muñoz Murillo – 1210061 Cristian David Munera Cardona – 1210060.

Object recognition with ultrasonic sensors [http://www.cs.cmu.edu/~motionplanning/papers/sbp_papers/integrated1/ihsan_object_rec_w_sonars.pdf]

Shape Recognition and Position Measurement of an Object Using an Ultrasonic Sensor Array [http://cdn.intechopen.com/pdfs-wm/37176.pdf]

SIMILAR JOBS

OBJECT RECOGNITION USING ULTRASONIC SENSOR