Interested in project "Application to measure sensor data and various parameters"

[Lakhshya Bansal]

Hi

I’m Lakhshya Bansal, male (have to mention it as people often misread the name), 19 years old. I’m currently pursuing my 4th semester (2nd year) of B.E.(Hons.) Computer Science at BITS Pilani Goa Campus. I’d love to work on the ‘Application to measure sensor data and various parameters’ project. I’ve been working on android for over 1 ½ years now and had picked up C/C++ programming in my 8th grade from NIIT. Some of my android work includes:

• Gingerbox Parental Control App: http://www.gingerbox.in/  This is one of my father’s startups to which I have contributed.

• WakeUp: https://github.com/lakhshya/WakeUp.git  An android based alarm app that will not stop until you wake up.

Firstly I’d like to mention that it’s an amazing idea considering the number of android smartphones today and the improvement of technology in every smartphone. Phones today are so powerful but a lot of that power is wasted by negligible usage. 

I feel there are two key approaches that we have to take towards the app. One is from the development point of view and the other more importantly is from the users’ point of view. Addressing the users’ point of view first:

• We need to identify the sensors and what use the data hence collected can serve.

• Allow analysis of multiple sensors simultaneously. E.g. Sound and gyroscope data can be analyzed together to verify the Doppler’s effect.

• We could add some games or tasks like the above verification of Doppler’s effect just for entertainment providing suggestions on what the data can be used for.

• Allow analysis of data collected regularly over a time-span e.g. a week.

• Export the data/graphs to memory so that the user can use the data on his PC and other devices as well.

With regard to the developer’s point of view:

• The app involves storage of good amounts of data which needs to be handled properly.

• Flexible GUI for different devices and screen sizes.

• Apt deallocation of sensor resources and services.

• Need to identify the appropriate graph library to be used in the app.

• Due to Android’s dynamics’ from phone to phone and version to version testing has to be done on a good number of android devices.

Please fill me in with your views on my understanding of the project and any advice that you have for me.

Thank you a ton for your time and this opportunity. It really means a lot to us students.

Best Regards

Lakhshya Bansal

B.E.(Hons.) Computer Science

BITS Pilani, K.K. Birla Goa Campus

Email ID : lakhs...@gmail.com

[Pankaj Nathani]

Hi Lakhshya,

Thanks for your introduction. 

 

• Gingerbox Parental Control App: http://www.gingerbox.in/  This is one of my father’s startups to which I have contributed.

I tried to look for a link to Google play store but couldn't find it on this page: http://www.gingerbox.in/Download.aspx

• We need to identify the sensors and what use the data hence collected can serve.
• Allow analysis of multiple sensors simultaneously. E.g. Sound and gyroscope data can be analyzed together to verify the Doppler’s effect.
• We could add some games or tasks like the above verification of Doppler’s effect just for entertainment providing suggestions on what the data can be used for.
• Allow analysis of data collected regularly over a time-span e.g. a week.
• Export the data/graphs to memory so that the user can use the data on his PC and other devices as well.

Your above understanding is mostly correct. You have an interesting idea of allowing analysis of data from multiple sensors simultaneously. Could you please elaborate on how will you achieve this? What kind of visualisations would you use for this specifically?

With regard to the developer’s point of view:
• The app involves storage of good amounts of data which needs to be handled properly.
• Flexible GUI for different devices and screen sizes.
• Apt deallocation of sensor resources and services.
• Need to identify the appropriate graph library to be used in the app.
• Due to Android’s dynamics’ from phone to phone and version to version testing has to be done on a good number of android devices.

Please fill me in with your views on my understanding of the project and any advice that you have for me.

 

Your understanding seems correct. We would like to have the application to function on smartphones and tablets since a large number of researchers may use this.

Please feel free to ask any specific questions that you may have about the application.

Regards,

Pankaj.

[Lakhshya Bansal]

Hi Pankaj Sir

Thank you for your reply. Sorry about the gingerbox link. The app can be found directly on the Google Play Store at https://play.google.com/store/apps/details?id=com.gingerbox&hl=en

In reply to how I would analyze data from multiple sensors at the same time:

By virtue of my findings till now, each sensor that the user would select will be registered to a SensorEventListener, thereby the app can independently receive data from each sensor. A SensorEventListener notifies you when there is a change in data and not on the basis of time though the gap between change of sensor state and data notification(broadcast) can be set based on desired precision and time lag. As regards the visualization, the data collected, if I am right will be a function of time. Thus line charts appear the most appropriate option of data representation other that tabulation, both of which should be stored in the secondary memory for future reference. We could superimpose the data from the different sensors or place them one above the other using the same X axis i.e. the time axis, with an appropriate factor of scaling for each variable. Allowing each sensor to update the graph could cause problems within the graph update, thus a separate thread/service could be used for the graph update that works at intervals of a few hundred milliseconds or as decided by the user.

These are the visualization I have thought of w.r.t. multiple sensor data analysis.

Scatter graphs could also be used for 2 parameter systems where one of the sensors would be along X axis and the other sensor would be along Y axis.

I have a few queries.

1. Are we focusing on only 2D visualization or will we involve 3D visualization too. 
2. Is the focus on collection of data so that it can be used on a more advanced system like a PC or do we intend for the data to be used and analyzed on the mobile device itself.
3. Is there any particular graphing library that you would recommend for the purpose of this project.
4. Do the mobile based projects have any connection to the BuildmLearn app

Please do correct me for any mistakes and kindly provide inputs on what I have missed out.

Thank you

Best regards

Lakhshya Bansal

[Lakhshya Bansal]

I think I need to rephrase my second question:

2. Is the focus on collection of data alone for analysis on more advanced systems like a PC or is the focus on collection of data and visualization on the mobile device itself thereby requiring more thought on how to present the data than on how to store the data.

[Pankaj Nathani]

Hi Lakhshya,

Thank you for your reply. Sorry about the gingerbox link. The app can be found directly on the Google Play Store at https://play.google.com/store/apps/details?id=com.gingerbox&hl=en

It works now :)

 

In reply to how I would analyze data from multiple sensors at the same time:

By virtue of my findings till now, each sensor that the user would select will be registered to a SensorEventListener, thereby the app can independently receive data from each sensor. A SensorEventListener notifies you when there is a change in data and not on the basis of time though the gap between change of sensor state and data notification(broadcast) can be set based on desired precision and time lag. As regards the visualization, the data collected, if I am right will be a function of time. Thus line charts appear the most appropriate option of data representation other that tabulation, both of which should be stored in the secondary memory for future reference. We could superimpose the data from the different sensors or place them one above the other using the same X axis i.e. the time axis, with an appropriate factor of scaling for each variable. Allowing each sensor to update the graph could cause problems within the graph update, thus a separate thread/service could be used for the graph update that works at intervals of a few hundred milliseconds or as decided by the user.

Thanks for the details. So you would present the user to select one or more sensors and plot them on the same graph / chart. Makes sense.

 Scatter graphs could also be used for 2 parameter systems where one of the sensors would be along X axis and the other sensor would be along Y axis.

I have a few queries.

1. Are we focusing on only 2D visualization or will we involve 3D visualization too. 

We are really looking forward to see some cool visualizations proposed as a part of this project. No reason we can't use 3D. 

1. Is the focus on collection of data so that it can be used on a more advanced system like a PC or do we intend for the data to be used and analyzed on the mobile device itself.

 Currently, we do not have the PC system in place to analyze the data. If you provide an option to export the sensor data to a file and somehow we could use it on the PC if required.

1. Is there any particular graphing library that you would recommend for the purpose of this project.

The project is going to be under new BSD license  so you are feel to use any graphics library compatible with this license.

1. Do the mobile based projects have any connection to the BuildmLearn app

No, the mobile applications are standalone applications. They are not related to the toolkit.

Regards,

Pankaj.