Friday, 2 September 2011

Motion Based Audience Response system - MARS

In this post I detail one of the recent developments at the School of Engineering at the University of Portsmouth, U.K., the all new Motion based audience response system (MARS), developed by one of my students Emeline.

Response systems are now widely used within classrooms and conferences. The carrier technology used to get the votes from the students to a centralised system is key in such systems. The faster the votes are collected the better the feel of the system. RF keypads perform well in this regards and votes are collected almost instantaneously (2-3 secs delay if that). SMS on the other hand can have longer delays. Tweets and emails can be within the 5 second window. But the truth is that depending on the class size any delay up to 10 seconds is acceptable as different students take time to respond to a question. 

In previous posts I have detailed some of our previous development of personal response system / Electronic voting systems / Audience response system (phew so many different names for the same thing!) based on SMS, Bluetooth, twitter and email. In this post I detail a new way to capture the response to multiple choice questions from a class of students using just a simple HD web cam and a laptop with some software that we developed.

Yes, its that simple!

No need for any clickers, keypads or mobile phones (with varying features) or wifi connection etc. Just a web cam suitable to capture your audience, a laptop (2GHz+, 3GB RAM+) and the shiny new MARS software.

Below is a sample screen-shot of how the system mixes a live view from a class-room with as many virtual keypads (with keys A, B, C and D) as there are students to be used by them to respond to a 4 choice question. Video is here now, see below:

Two types of virtual keypad - Horizontal and Square.
The image you see above is what the student see on the over-head projector, live, as the camera mounted in front of them captures these frames and mixes the black and white virtual keypads wherever it detects human motion. The system can provide these virtual keypads for around 70 students in the audience (limited by the size of the keypad and camera resolution). 

Now, imagine the students can also see on the overhead projected screen a question with four choices, one for each letter in the virtual keypad. As students have individual keypads they reach out to the option of their choice with their hand and select the response they want to answer the current question. This is shown in the image below. 

This is how the overhead projector screen looks like in full.
Example class using the system

As you see in the picture above the individual options (A, B, C and D) change colour when students hover their hand over the individual boxes. To vote a student simply hovers their hand three times over the option of choice and the colour changes from orange to red and finally to green (yes, not quite the traffic light!). When the teacher sees enough green lights then they can stop and collect the stats, as shown in the image bottom right.

Reflections after 1st use with students:
I will continue trialing the system in schools (two planned) and in my own classes (as you can see in the pictures above) so more details to follow. Today (19th Sep 2011) I used this system first time with my students and their immediate reaction was "Cool", "Can we have another go at the system" (after I stopped it)" and "it would be nice to get the virtual keypad not to overlap my eyes". The last response can easily addressed when allocating boxes manually as opposed to based on motion. The session lasted 30 minutes and i was able to ask 6 questions and talk about the results and other things within the time. All responses were in within 1 minute.

On the issue of shy students - I think, and I still need to evaluate it properly, the faces of people are covered with the virtual keypads (see pics above) and within 1 min students will probably focus in getting their response in as opposed to check out a) who is sitting where and b) find out what their response is. After the minute we only show the stats and the votes are cleared.  

Any thoughts, suggestions, applications and inquiries are welcome. Like this post all comments will be treated as creative commons license relevant to this blog (i.e CC BY-NC-ND 2.0).

Happy teaching to you all.


@polarisdocta said...

Data-collecting power of #Clickers w simplicity of cards. Anonymity lost, tho: MARS by @manmalik (via @courosa)

@chrismattos said...


Martine Hawksy (@mhawksy) said...

In several tweets, Martin said...

It would be interesting if there was an easy way for people to test the s/ware eg click here to reg for closed beta and download.

Its interesting the way everyone votes at the same time wonder how lack of anonymity will change the dynamics.

Could there be a blind version using colored card that doesn't need to display audience Cam on screen?

Edublend said...

Thanks to all the comments on twitter and here and the +1s/likes on G+/FB.

@polarisdocta, we do not use any cards, just special hand movements to vote. The anonymity is lost in the sense that students can see what others have votes but that is only for a short span of time usually 10-15 seconds when everyone has votes in the audience. Once data is collected the data is anonymous as we cannot detect in this version who is who!

@Martin, thanks for your tweets. The aim for this development was to realise a system that does not rely on any thing to be taken to classrooms, distributed and collected back (clickers, colored sheets etc). Anonymity can be addressed in other ways.

I am interested to see if the loss of anonymity will make any difference in which people vote with this system as it will all happen in 10-15 seconds for a large class of 70.

Edublend said...

Welcome back if you are returning to this page.

I added some pictures and some reflections from today's use of the system with my class, see last few paragraphs above.


Edublend said...

Coming Soon, a trail version for you to try out at your end.

Dragos Ciobanu said...

On twitter, Dragos Ciobanu
@suchadrag wrote:

@manmalik what a cool idea!I agree that anonymity is lost&maybe circular rooms will be harder2manage,but liking it! I'll ask around,thanks:)

Edublend said...

Hi Dragos,

The shape of the room does not matter as long as the audience can fit within the view angle of the camera. The boxes that you see are a result of head detection, so this will work any where.

Regarding anonymity, having though about it long and hard, my system does not record who voted what, but as I said that can be done. Therefore currently this system is more anonymous than say clicker systems that record who voted what for the lecturer to see.

And anonymity during voting , as you can see from some of the images, the face of the person is not visible as its mostly hidden behind the keypad. If there was a really shy student, they can choose not to vote at all or just pretend to be voting and never actually vote (green stage). For the not so shy students this will even start discussions on why they chose a given option.

As I said we can do the visually anonymous voting but it will require handing out A4 sheets with QR / Glif codes. Staff could then track who voted what and even use it for attendance. This blog post describe a system that requires nothing but the software and a camera to get votes in.

Today I used it with a class of 100+ students and only 50 could take part (using only one camera) but the others were given a paper to mark their answers and take part. However I was able to setup in 5 mins and in 10 mins i was 4 questions down my list of questions.

Dr. Toby Howard said...

Many thanks Manish. Your stuff looks great!

We are experimenting with clickers, and my research group
also has interests in image processing tracking etc.

best wishes