IoT Sensors & Integration with Physical World
Download as PPTX, PDF0 likes70 views
The document provides an overview of various types of sensors and their integration with the physical world. It covers definitions, classifications, examples (such as accelerometers and MEMS sensors), and the importance of sensor fusion and data processing in deriving actionable insights from raw data. Practical applications and the significance of sensor data modeling and filtering are also discussed.
IoT Sensors & Integration with Physical World
- 1. IoT Sensors & Integration with Physical World 1
- 2. What is Sensing ? Collect information about the world Sensor - an electrical/mechanical/chemical device that maps an environmental attribute to a quantitative measurement ◦ attribute mixtures ◦ hidden state in environment Each sensor is based on a transduction principle - conversion of energy from one form to another Also known as transducers 2
- 3. Where is the cropline? Autonomous harvesting Sensing for specific tasks 3
- 4. Type of sensors Active ◦ send signal into environment and measure interaction of signal w/ environment ◦ e.g. radar, sonar Passive ◦ record signals already present in environment ◦ e.g. video cameras 4
- 5. Type of sensors • Classification by medium used: ◦ based on electromagnetic radiation of various wavelengths ◦ vibrations in a medium ◦ concentration of chemicals in environment ◦ by physical contact 5
- 6. Some Practical Examples of Sensors with their working principals 6
- 7. Accelerometers sensors 7 • Acceleration is a measure of a change in velocity (change of speed or direction). • The parameter being measured may be a static force, such as gravity exerted on a device. Other sensors make dynamic force measurements to measure motion changes and vibration. • An example of an acceleration sensor is a chip in a moving vehicle that measures changes of speed and uses high acceleration readings (such as during an accident) to trigger an airbag to protect the passengers. • In some industrial applications, the vibrations detected by an acceleration sensor could be an excellent indicator of a potential problem with a moving part—such as a motor with bearings that are worn.
- 8. MEMS sensors 8 • Inertial measurement units (IMUs) are found in every smart watch and fitness tracker, making the most of mature MEMS components that are reliable, familiar and cheap. • In modern, high-end smartphones, integrated chip sensors to measure motion, direction, pressure, magnetic fields, etc., are becoming quite common. • In chip form, these are usually Micro-Electro-Mechanical Systems (MEMS) sensors for many different parameter measurements. • The implementation of MEMS uses ultra- miniaturized physical structures—beams, arms, and associated electronics—to measure the motion of the structures when the chips moved. The device essentially converts a mechanical (physical) motion into an electrical signal. • A gyro sensor, for example, senses rotational motion and changes in orientation. These can be used in a variety of applications, such as correcting for hand-held shake in video and still-image cameras and human motion sensing for video games.
- 9. Bend sensor 9 a variable resistor resistance changes as it bends V = I x R assuming constant current, the measured voltage changes with resistance
- 10. Sample problem 10 Bend sensor specs: 100 when straight 1000 when bent Given a 5 V source, what is the min. and max. current that is drawn? V = I x R I = V R min = 5 1000 = 5 mA max = 5 100 = 50 mA
- 11. Sound sensor 11 • SONAR: Sound Navigation and Ranging – bounce sound off of something – measure time for reflection to be heard - gives a range measurement – measure change in frequency - gives the relative speed of the object (Doppler effect) – bats and dolphins use it with amazing results – robots use it w/ less than amazing results
- 12. Odor sensor 12 • Detection of chemical compounds and their density in an area – spectroscopy - mostly lab restricted – chemical detection - sniffers and electronic noses via “wet chemistry on a chip” • No major penetration in robotics yet applications are vast (e.g. mine detection)
- 14. Pressure sensor • A pressure sensor usually acts as a transducer; it generates a signal as a function of the pressure imposed. • Pressure sensors are used for control and monitoring in thousands of everyday applications. Pressure sensors can also be used to indirectly measure other variables such as fluid/gas flow, water level etc. • Pressure sensors are used for a range of diverse applications in industries such as medical, aerospace, transportation, and general industry and for test and measurement. 14
- 15. Piezoelectric pressure sensor 15 • Uses the piezoelectric effect(it is the electric charge that accumulates in certain solid materials in response to applied mechanical stress) in certain materials such as quartz to measure the strain due to pressure. • Used to measure high dynamic pressures.
- 17. Moisture sensor • The moisture sensor works when the series of discs in the sensor expands they get wet, and to a certain size, they flip a micro switch. • This interrupts the current going from the controller to the valves, whereby the controller will tell the valves to start watering in the middle of a rainstorm, if the valves do not get the memo then they don't turn on. • When the discs dries out in the sun and the wind, then the switch flips back, and the valves will water when the controller tells them to start watering. 17
- 18. Sensorfusion 18 Our daily challenge here is in turning raw data into useful, or 'actionable' data. Sensor fusion is the process of combining sensor outputs from multiple sensors to gain greater total insight. The most common example is using individual xyz acceleration and rotation data (e.g. from a 6-axis IMU) into motion data. This in turn can be used to count steps, differentiate between activity types, and so on.
- 19. Sensor choice 19 • What sensors to employ ? • e.g. mapping – ranging - laser, sonar, IR, stereo camera pair – salient feature detection - doors using color • Factors – accuracy, cost, information needed etc. – Where do you put them ?
- 20. Conversion to digital data 20 • Analog-to-digital conversion is an electronic process in which a continuously variable (analog) signal is changed, without altering its essential content, into a multi-level (digital) signal. • Sensors are often used in local applications, where their signal is processed using circuitry designed for that local application. However, in a sensor that is used for remote data transmission of the measurement, the electrical signal must be converted into a digital value, or number, for the transmission. • The specific electrical signal from different sensors may vary over a wide range of current or voltage or other electrical parameters (such as resistance or capacitance) and often must be converted and amplified into a voltage that can be measured.
- 21. Sensor data filtering • If necessary, the signal must be filtered to eliminate noise or to reduce the frequency of the measurement for the requirements of the application. • For example, a temperature sensor generally changes its value relatively slowly as the sensor matches its environment. Therefore, a rapid change in reported temperature may be an inaccurate reading, which should be filtered to reduce potential errors. 21
- 22. Sensor data sample 22 Data Source: https://www.kaggle.com/nphantawee/pump-sensor-data
- 23. Sensor data modelling • Data Pre-processing: ◦ Data Cleaning – Ignore values, Fill missing values. 23
- 24. • Data Transformation: ◦ Normalization: It is done in order to scale the data values in a specified range (-1.0 to 1.0 or 0.0 to 1.0). ◦ Attribute Selection: In this strategy, new attributes are constructed from the given set of attributes. ◦ Discretization: This is done to replace the raw values of numeric attribute by interval levels or conceptual levels. 24 Sensor data modelling
- 25. • Data Reduction: ◦ Attribute Subset Selection : highly relevant attributes should be used, rest all can be discarded. ◦ Dimensionality Reduction : Wavelet transforms and PCA (Principal Component Analysis). 25 Sensor data modelling
- 26. Predictive modeling • Once the data pre-processing, Transformation, Reduction is completed. • We can use Machine Learning techniques to model our data into a predictive function. • Either it could be Supervised (classification or Regression) or Unsupervised (Clustering), depends upon the problem we are trying to solve. 26
- 27. Thank You 27