AN APPROACH ON - SENSOR DATA
ACQUISITION USING MEMS
.
Abstract
Ergotic – to manipulate the environment.
We present the ongoing work on a project for gesture Epistemic – to discover the environment through tactile
recognition .Considerable effort has been made on this experience.
project to provide a novel interface for Human Machine
Interaction(HMI) that can be used to give alphabets as Gesture recognition is the process by which gestures made by
the user are recognized by the system. Despite the richness and
input to internet browser in an interactive set top box. The
complexity of gestured communication, researchers have made
device is designed with a MEMS sensor which is rubbed progress in beginning to understand and describe the nature of
on the tabletop using our hand to send alphabetical input gesture. Kendon [2] described a “gesture continuum” defining
to an interactive set top box .Users write English alphabets five different kinds of gestures:
in capital letter by just using the sensor. Then this data
undergoes some filtering and is finally sent to the Gesticulation – Spontaneous movements of the hands
recognition module .Final decision about the recognized and arms that accompany speech
character is made with a recognition accuracy of more Language-like gestures – Gesticulation that is
than 90.96%.In this paper we discuss about the method of integrated into a spoken utterance, replacing a particular
data acquisition from the MEMS.We have tested the spoken word or phrase
Pantomimes – Gestures that depict objects or actions,
system with the inputs given by 22 users (16 male and 6
with or without accompanying speech
female candidates) and each of the user gives the input 5 Emblems – Familiar gestures such as “V for victory”,
times. “thumbs up”, and assorted rude gestures (these are
Often culturally specific)
Index terms – Gesture recognition, MEMS, Sensor. Sign languages – Linguistic systems, such as American
Sign Language, which are well defined.
In the paper [6], Raymond et al recognizes the characters written
I. Introduction in American sign language and a modified version of it. But the
problem of this approach is that a user needs to memorize a lot
of code/symbol for each character. But in our approach we give
In this section we try to analyze the concepts of gesture the alphabets as inputs which are not difficult to remember.
recognition and the previous discoveries made in this field. Lee and Yangsheng Xu et al uses a gesture based approach to
Gesture recognition is an area of active current research. The make a system which can not only interact with the user to
prospect of user-interface in which natural gestures can be used recognize the gestures but also learn several new gestures and
to enhance human-machine interaction bring visions of more update its knowledge of gestures. The commonly used
accessible computer systems, and ultimately of higher approaches of gestures are HMM based approaches, .
bandwidth interactions than will be possible using keyboard and These above mentioned approaches are not able to recognize the
mouse alone. Gestures are expressive, meaningful body motions characters; they are limited to the recognition of gestures. But
– i.e., physical movements of the fingers, hands, arms, head, we give character input to any internet browser which needs to
face, or body with the intent to convey information or interact be recognized and for this we formulate some of the approaches.
with the environment. In this paper we discuss the process of sending the character
inputs to the recognition module including the preliminaries of
Earlier there were mainly three functional roles of human data acquisition from the MEMS.
gesture that was described:
Semiotic – to communicate meaningful information.
II. Overview of sensor data The overall process of data acquisition and data sending is
acquisition and data depicted in Error! Reference source not found.Error!
Reference source not found. and Error! Reference source
transmission not found.Error! Reference source not found. respectively.
Start
The data acquisition unit (DAU) aims at providing a hand held
wireless device, a sensor that utilizes the technology of Micro-
Electro-Mechanical System (MEMS). The accelerometer type
MEMS chip is used to sense movement of the device. We
measure the acceleration and then compute the displacement of
the device. The DAU consists of the following modules:
If there is any
movement in
Hand held device: The block diagram of this device is
MEMS sensor
shown in the Figure1. The hand held device consists of No
the MEMS sensor, a transmitter RF- SoC (Radio
Frequency- System- on Chip) and a microcontroller.
Receiver apparatus: The block diagram of this device is Yes
shown in the Figure2.The receiver apparatus
Read Acceleration values from
compromises of a microcontroller, receiver RF-SoC
MEMS sensor
and the USB interface which is connected to the Set
Top Box (STB).
The accelerometer based MEMS consists of a LIS302DL chip, Send acceleration values to RF SoC
the micro control unit (MCU) is MSP 430 which is a 16 bit to transmit
processor. The transmitter transmits at 2.4 GHz and at the
receiver side the RF receiver is also tuned at the same frequency.
The MSP 430 consists of a chip called as F 2274 and at the
receiver side this is connected to the USB interface.
The USB interface will then be connected to the Set Top Box Figure 3 : Data Acquisition Unit Flowchart.
(STP).
Start
Accelero Micro RF
meter Control Transmitter
If there is any data
based Unit CC 2500
on RF Receiver
MEMS MSP 430 2.4 GHz
sensor No
Figure 1 : Data Acquisition Unit. Yes
Read data from RF Soc
RF Micro Host Set Convert acceleration value into
Receiver Control (USB Top displacement
CC 2500 Unit interface) Box
MSP 430
F2274 Send calculated data to host
Figure 2 : Data Receiving Unit.
Figure 4 : Data Receiver Unit Flowchart. Figure 5(b): Schematic diagram of the Evaluation
board.
III. Details of Evaluation board
and Evaluation kit EVALUATION KIT
The EK302DL is an evaluation kit designed to provide the user
Initially we used the MEMS sensor and an application kit with a complete, ready-to use platform for evaluation of the
provided by ST Microelectronics to record the values of LIS302DL.
acceleration that indicated the movement of the sensor. In
addition to the MEMS sensor, the evaluation board utilizes an The installation of the Graphical User Interface (GUI) for the
ST7-USB microcontroller which functions as a bridge between EK302DL requires two steps:
the sensor and the PC.
1) Installation on the PC of the software delivered with the
The sensor chip used is LIS302DL.It gets connected to the evaluation kit.
computer via the cable in an USB port. The LIS302DL is a low 2) Installation of the Virtual COM driver needed to use the
power 3-axis linear accelerometer with digital output. The evaluation kit board
device includes a sensing element and an IC interface capable of
translating information from the sensing element into a Once the installation is complete, a COM port number will be
measured signal that can be used for external applications. assigned to the ST Virtual COM driver.This number should be
retained as it will be required to run the EK302DL Evaluation
Software GUI. In our applications we have chosen port number
#3. This port number was also used when we run the GUI.
Both the hardware and software that compose the EK302DL
Evaluation Kit have been designed to operate with Microsoft
Windows XP.
To execute the EK302DL Evaluation Software GUI:
1) Click on Start > All Programs;
2) Select EK302DL > Executables;
3) Launch the program “EK302DL Ver.1.3”.
We can roughly divide the evaluation kit GUI interface into 3
different regions. One region shows whether the COM is
connected or not and gives the COM number, the second region
tells us the name of the file in which the recorded information
should be saved and the third region shows how the information
gets recorded and what are the different values recorded at any
instant.
The initial information obtained is the acceleration information
which is later converted to velocity and displacement
information.
The evaluation kit is shown in the following figure that helps us
to record the average acceleration values in the x, y and z co-
ordinate. Then we try to extract the useful information as per our
requirement.
Figure 5(a): Evaluation board that contains the
LIS302DL chip.
“Angle” - Returns the tilt angle, expressed in degrees,
that is inferred from the “ADC Out” data.
The ADC output value that displays the acceleration data
provided by the sensor in all the three different co-ordinates: x
co-ordinate, y co-ordinate, z co-ordinate.
Figure 7 : X, Y and Z axes taken as reference
while using the sensor.
Initially for our work we assume that the two axes x and y are
taken on the plane (that may be the table top on which the sensor
is kept) and the z axis is perpendicular to the plane. The diagram
above shows the orientation of the three axes with respect to the
sensor.
IV. Details of data acquisition
We are getting the data from the sensor. Users write English
alphabets in capital letters by just moving the sensor on the
table. We take the input for about 110 users (16 male and 6
Figure 6 : Evaluation software GUI. female where each writing the 26 alphabets 5 times). The input
data comes as a stream of x, y and z co-ordinates that gives the
acceleration values. At the real time we are able to read 2048
data at a time. The device captures the data and stores the data in
“DATA” TAB a buffer. For each point the x, y and z coordinates are stored in a
packet. Now these packetized information is sent from the data
The Data tab shows the acceleration values measured by sending unit to the recognition module. Now for processing we
LIS302DL sensor. It is divided into three boxes: would read the continuous data that is stored in the buffer.
“ADC Out” - Displays the acceleration data provided
by the sensor after its conversions from 2’s complement
to magnitude and sign. V. Details of data processing
“Acceleration Value” - Represents the acceleration data
measured by the sensor, expressed in mg.
Then some preprocessing is done on the data values so that the
characters can be recognized in the recognition module. After
recording the acceleration values we analyze the data values. We
used our simple algorithms to convert these acceleration values
to velocity and position coordinates. Now we try to analyze the
behaviour in different domains - acceleration domain, velocity
domain and displacement domain.
On thorough analysis we have derived to the following
conclusion.
The vz–coordinate (z coordinate in the velocity domain) gives us
the information about the no of strokes in the alphabet. As the Figure 9 : Vz values of the alphabet K
user lifts his hand to take each stroke there is a sudden variation
in the z-coordinate. Thus if we plot the z values we get a peak in
the graph at the point where the user lifts his hand. There may be The Figure10 below shows that we need to take three different
several peaks if we are plotting the raw values. So we do an strokes for writing the alphabet “A” and only one stroke for
initial median filtering on the vz values. Then we plot the datas writing the alphabet “Z”. The arrows in the figure show the
and try to identify the strokes. number of strokes.
Figure 8 : Vz values of a character Figure 10 : Strokes in the character "A" and "Z"
By analysis of the above figure we can find that there is no sharp
peak availability in the graph and hence there is no lifting of
hand during the writing of the character. Hence we can infer that
there was only one stroke taken for writing the character. Thus
by using this process we can separate the no of strokes in each
character and obtain the corresponding coordinate values of the
strokes in each character.
The Figure 9 below shows the vz curve for the alphabet “K”. In
the graph we find that there is one sharp peak which indicates
that there is one single lifting of the hand from the table top,
hence there are two strokes taken while writing the alphabet “K”
whereas three different strokes are required to write the alphabet
“A” and in order to write “Z” the user needs to take only one
stroke.