gender survey for robotics

[Admin]

It has come to my understanding that women comprise only about 6% of all robot builders (according to The Society of Robotics and Automation).

As a first step towards attempts to rectify this problem, I am interested in seeing if this is true for the general robotics hobbyist so I posted this simple poll.

Why do I care? Well, the more robot builders we have on SoR, the more people we have sharing ideas and helping beginners. If RAS is correct, we could potentially double our member base . . .

[Tsukubadaisei]

Interesting, last year they did a survey in all robotics/mechatronics departments at my university and found that 20% of the members(including teachers, undergraduate, graduate and doctor students) were women(the total is about 400 people which results in about 80 girls). I thought I had bad luck....

[polar bear6]

its official, SoR is a sausage party!

[Admin]

looking at the forum stats, apparently the newest version of the forum I installed now does gender stats:

Male to Female Ratio:     22.5:1

if my math is right, thats like 4.44444444444% women . . .

But of course this percentage can be scewed because members arent required to say whether they are male or female during registration.

so the next question is . . . how could this problem be solved?

[Nyx]

Is there really a problem to solve to start with? And can you really do something about it, seeing that in general, there are less female robot builders.

My personal advice, if you want to attract more females, is to add more theoretical material. In my opinion, your site (no offense intended) is overly focused on low-level electronics and not enough on algorithms and more advanced robotics (past the hobby/toy stage).

[Admin]

In my opinion, your site (no offense intended) is overly focused on low-level electronics and not enough on algorithms and more advanced robotics (past the hobby/toy stage).

any specific requests?

and dont say AI, cause its not a solved problem yet

[Nyx]

Having an Artificial Intelliigence, or at least a Programming forum would be a good start. Perhaps links to various AI tutorials, and more focus on more advanced problems/solutions. I think that topics like neural networks, genetic programming, clustering, classification are all very useful for robotics, and deserve some treatment.

I don't know if you're allowed to host the papers, but if not, there are tons of robotics papers that would be nice to have links to... Examples:

http://people.csail.mit.edu/brooks/papers/AIM-864.pdf
http://www.cse.unsw.edu.au/%7Eclaude/research/papers/ML9.pdf
http://www.cim.mcgill.ca/~jer/courses/ai/readings/Mahadevan_Automatic.pdf
http://www.ri.cmu.edu/pub_files/pub3/pomerleau_dean_1991_1/pomerleau_dean_1991_1.pdf
http://gral.ip.rm.cnr.it/nolfi/papers/nolfi.perac.pdf

Maybe I can write you a few tutorials once I have completed my robot.

[Tsukubadaisei]

I partially agree with Nyx. However I think I know a better way to attract women to robotics. Half of my robotics team are women. I think that cute robots receive more attention. Every time I am looking for girls for my team, I present a new project. I name the robot the name of a girl, a make it cute, I try try to avoid manly ways of solve problems and so on. I try to bring balance. Another thing I observed is that women USUALLY pay more attention to details (well, I am a man and details are a priority). The tutorials in this page are a bit superficial. They are very good for newcomers but but when you really want to do some thing beyond they are useless(sorry for the word, no offence at all)

Right now I participating in 3 contests (including ADMIN's one) and I will be using the same robot for all of them. When I post my tutorial it will be clear what I mean.(the robot I am building is pretty complex so the tutorial may not be the best but will be the longest and most detailed ever, I promise)

Now about Nyx ideas, they are very good and I would love to have all those links. But some of them, like neural networks and genetic programming are far too advance for hobbyists like most of the users here, and even for pros. Those subjects are still open problems under research in hundreds of labs around the globe.

PS: Of course, being sexy like me helps attract women as well...

[Admin]

The tutorials in this page are a bit superficial. They are very good for newcomers but but when you really want to do some thing beyond they are useless(sorry for the word, no offence at all)

Explain! What should I do differently in my tutorials?

I assume that what you mean is my theory pages dont go down to the step-by-step how-to level? I strongly feel that if someone has a good background in theory, they can figure out the details on their own.

For example, suppose I write up a detailed how-to for ATmega8 microcontrollers without theory. What happens if you dont want that microcontroller, but instead want an ATmega168? What happens then? I cant write a tutorial for every single microcontroller out there . . . this is why it needs to be 'superficial.'

My robots page has detailed examples, with code and designs and everything you need to copy. When I get time Im gonna revamp a bunch of the tutorials with new stuff - many of the older ones were quickly written placeholders.

My site is also very new, not even 2 years old. Over time I hope to develop enough tutorials to cover all the details in full. Just let me know what needs to be done differently to better help you and Ill do my best!

And of course, if you havent already, you can have influence on what I write by voting on the next tutorial:
http://www.societyofrobots.com/robotforum/index.php?topic=7.0
There are currently a few advanced topics in the list, such as pathfinding.

Perhaps links to various AI tutorials, and more focus on more advanced problems/solutions. I think that topics like neural networks, genetic programming, clustering, classification are all very useful for robotics, and deserve some treatment.

I consider myself a total noob with these . . . I understand them but never implemented them. I do keep a list of advanced robotics papers in the main menu. If you find papers that are really useful just send em my way and Ill add them on. Survey papers I find the best, so DEFINITELY send/read those!
When I get time Ill read through the links you already sent . . . busy busy busy!

at least a Programming forum would be a good start

doesnt this count?
http://www.societyofrobots.com/robotforum/index.php?board=2.0

I name the robot the name of a girl, a make it cute, I try try to avoid manly ways of solve problems and so on.

lets see . . . cute names of my robots:
Pikachu (I have made about ~8+ robots with this name)
Hyper Squirrel
Stampy
Fuzzy
Carpet Monkey
Cheezy Poof

If I ever make a battlebot, Id probably call it 'Squishy' or 'Tiny' . . .
But I might draw the line before painting it pink . . .

[Nyx]

I think one of the big problem in robotics, in general, is that there tends to be a disjunction between the "robot builder" (engineering) community, and the AI community. I don't just mean here, but all over the web... And even at my university. The AI people work on some very narrow AI problems, using some highly reductive simulations... And on the next floor up, the engineering people work on building robots for special purposes... They do get together sometimes, but definitely not often enough. There are more girls in the AI (CS) part, less in the engineering dept. Like I said above, I think you should consider having an AI forum, or at the very least a forum dedicated to programming (not the same thing as software in general).

As far as you being a "noob" about those things, or those problems being open ended... Alot of nice things have been done using some very simple approaches... See the paper about ALVINN, the self-driving car... The design of its neural network is extremely simple, and has been extended to driving around in a city environment, apparently. The ideas behind genetic programming, neural networks, etc, are usually not that complex... Fairly easy to pick-up... And many papers basically give you recipes to replicate their results (make your own car that drives itself!).

I thin what Tsukubadaisei meant is that your tutorials are overly specialized. You're not giving enough say, theoretical background, or examining the problem from a wide enough angle, so that the ideas can easily be applied to more complex projects. I personally think that this can be fixed by adding some more theoretical tutorials, instead of trying to modify all the existing ones, making them potentially more confusing to newcomers. Anyways, I know you might not think it should be your job to educate people about the more theoretical aspects of robotics (hey, people can just learn that in university)... But sometimes academics are impressively clueless about what they are doing, and so providing some public access to robotics knowledge could help alot of people.

To sum my point... I think you should shift the focus from "How to build a robot tutorial" (in reference to your webpage title) to "Making robots", "Creating robots", or a little bit of both.

[Admin]

I think one of the big problem in robotics, in general, is that there tends to be a disjunction between the "robot builder" (engineering) community, and the AI community. I don't just mean here, but all over the web... And even at my university. The AI people work on some very narrow AI problems, using some highly reductive simulations... And on the next floor up, the engineering people work on building robots for special purposes... They do get together sometimes, but definitely not often enough. There are more girls in the AI (CS) part, less in the engineering dept.

I 100% agree. Its especially a problem at CMU . . .

a forum dedicated to programming (not the same thing as software in general).

I dont quite understand . . . the software section was meant for all forms of programming . . . at the moment there isnt enough posts on AI to reference a new category. this post will explain why im resistant to adding more categories:
http://www.societyofrobots.com/robotforum/index.php?topic=7.msg3807#msg3807

The ideas behind genetic programming, neural networks, etc, are usually not that complex... Fairly easy to pick-up... And many papers basically give you recipes to replicate their results (make your own car that drives itself!).

I agree. I usually avoided these things because Im not a strong programmer. I understand all the background theory, but Im sure its much harder 'in the details.' I also have a strong preference to using microcontrollers, but they simply cant handle these types of algorithms . . . My AI experience doesnt really go past the really basic Fuzzy Logic . . . Sooner or later I will take your advice on this . . .

I thin what Tsukubadaisei meant is that your tutorials are overly specialized. You're not giving enough say, theoretical background, or examining the problem from a wide enough angle, so that the ideas can easily be applied to more complex projects.

I totally didnt see this coming . . . Ok so SoR used to get a lot of complaints about not being specialized enough. For example, check the hundreds of posts on computer vision - they dont just want theory but want specialized instructions . . . The reason for the $50 robot was because dozens of people complained about not having a detailed step-by-step tutorial . . .

But I agree with both you and them. They need a strong theoretical background, but also specific examples just to get people started . . . I try to do both . . .

To sum my point... I think you should shift the focus from "How to build a robot tutorial" (in reference to your webpage title) to "Making robots", "Creating robots", or a little bit of both.

I think most of my older tutorials covered the specialized tasks, and my newer ones will be more theory from now on. From now on Ill include additional theory with my future tutorials, and implement the theory using the $50 robot so anyone can use my source code right away. Is this what you guys want?

Probably part of the problem is that Im just one person trying to write about the entire field of robots to people ranging from complete novices to experts

[Nyx]

I think most of my older tutorials covered the specialized tasks, and my newer ones will be more theory from now on. From now on Ill include additional theory with my future tutorials, and implement the theory using the $50 robot so anyone can use my source code right away. Is this what you guys want?

Sounds like that could interest many people.

Once my robot platform is working... Perhaps I can make a tutorial about that, if you're interested. It's more like a $1000 robot, but it will be able to run any kind of algorithm and do extensive vision processing. My mom made a nice multi-level wooden base for it I'm still in the process of buying the missing parts though.

I could also perhaps write some basic general learning AI tutorials if that is of interest to people.

[dunk]

I totally didnt see this coming . . . Ok so SoR used to get a lot of complaints about not being specialized enough. For example, check the hundreds of posts on computer vision - they dont just want theory but want specialized instructions . . . The reason for the $50 robot was because dozens of people complained about not having a detailed step-by-step tutorial . . .

for what it's worth, whenever i'm researching robotics on line it's practical examples i'm looking for.
i want circuit diagrams and pre written code whenever possible.
maybe it's just me but i like to learn by example. i take the bits i need and add them to my knowledge.

while the SOR tutorials are all very good for explaining the theory, only the $50 robot shows you how to put the components together.

i think what would be good from a teaching point of view is for simple additions to the $50 bot so people can learn how to add one component at a time now they have a working platform.

maybe add a flashing LEDs to the bot for a bit of bling.
explain how to read the datasheet to choose I/O pins, set the pins to output, simple code snippet to switch on and off the LEDs, etc.

next maybe a bump sensor so the robot can backup of it crashes into a switched on torch.

this way people can learn how to program single I/O pins rather than just using your pre written code.
hmm. your probably way ahead of me here.

Probably part of the problem is that Im just one person trying to write about the entire field of robots to people ranging from complete novices to experts

so you shouldn't have to do it all yourself. i'm sure you are busy enough fighting spam bots.

i know i have been promising to write up some more tutorials for ages now. (my first one was a little obscure and probably not that usefull to most people. http://www.societyofrobots.com/member_tutorials/node/25)
i promise my next one will be more usefull.

what about all you $50 bot builders? does anyone have any cool modifications happening yet?
what about different firmware? anyone made their bot look for a dark place to hide? (the $50 cockroach...)
post your modifications up on the forum.
help Admin with suggestions on how to improve his tutorial.
earn cool points by being the first to post a pic of your $50 bot. (100 cool points = 1 theoretical chocolate bar.)

as for us more experienced bot builders, tutorials that i think would be usefull are basic examples of the building blocks people need to make their own projects.
i recently moved from developing on PIC to AVR microcontrollers.
on line examples would have made the step far easier.
basic tutorials i think people need for PICs and AVRs (in no particular order):
    "Hello World" flashing LED program and circuit.
    basic I/O. (including MOSFET example for driving higher loads.)
    UART communication (to controll your microcontroller from a PC).
    how to use a timer.
    how to use an interupt.
any others?

dunk.
(who will document some more of his projects soon.)

[Admin]

Once my robot platform is working... Perhaps I can make a tutorial about that, if you're interested. It's more like a $1000 robot, but it will be able to run any kind of algorithm and do extensive vision processing.

Im definitely interested! I think quite a few other people who want vision on their robots are also interested
I believe Somchaya will be working on a similar tutorial?

hmm. your probably way ahead of me here.

although not the original goal of the project, this definitely occurred to me ~3 weeks ago
I have plans to put my 'Stampy technology' on the $50 robot as the very next project
Now everyone can have a robot that beats up their cat!

Quote

Probably part of the problem is that Im just one person trying to write about the entire field of robots to people ranging from complete novices to experts

so you shouldn't have to do it all yourself.

I figure Ill just get the ball rolling, and in another year or two the beginners will start to become intermediates, meaning they will invent new ideas to share too

basic tutorials i think people need for PICs and AVRs (in no particular order):
    "Hello World" flashing LED program and circuit.
    basic I/O. (including MOSFET example for driving higher loads.)
    UART communication (to controll your microcontroller from a PC).
    how to use a timer.
    how to use an interupt.
any others?

My latest robot requires uart stuff so Ill be adding that to the $50 robot soon. Same with the LED hello world. The others however I have no plans in the near future to write about . . . and if you check the vote for a tutorial page, there are tons more ideas . . .

[ed1380]

I know it is mostly constructive criticism, but I'm kinda upset at people complaining about Admin's hard work, and all the great tutorials he has written for us. If you don't like the way he wrote the tutorial then modify it and get admin to post it up, so there will be many views on the same subject.


BTW I can't do nothing to my $50 bot since I can't program it and one of the sevos is messed up

[Admin]

I know it is mostly constructive criticism, but I'm kinda upset at people complaining about Admin's hard work, and all the great tutorials he has written for us. If you don't like the way he wrote the tutorial then modify it and get admin to post it up, so there will be many views on the same subject.

thanks! as Ed said, specific examples of how to improve a specific tutorial will be the most effective.

But I still encourage criticism, even if I sometimes have trouble taking it . . . to be honest I often tend to be the stubborn type . . . but if I cant take criticism, Im blocking myself from changing and improving . . . so bring it on!

Anyway, I often use the confusion within the forum on a specific subject to clue me in that a tutorial wasnt good enough . . . Ive updated/revamped probably a 3rd of my tutorials because of forum questions . . .

[Somchaya]

Once my robot platform is working... Perhaps I can make a tutorial about that, if you're interested. It's more like a $1000 robot, but it will be able to run any kind of algorithm and do extensive vision processing.

Im definitely interested! I think quite a few other people who want vision on their robots are also interested
I believe Somchaya will be working on a similar tutorial?

Yep, I'm working on that. I'm writing up a tutorial that describes how I did Whisk, essentially putting a laptop on wheels and doing vision and all that. I'll probably include some of the vision algorithms I played around with and how they work.

basic tutorials i think people need for PICs and AVRs (in no particular order):
    "Hello World" flashing LED program and circuit.
    basic I/O. (including MOSFET example for driving higher loads.)
    UART communication (to controll your microcontroller from a PC).
    how to use a timer.
    how to use an interupt.
any others?

So my next project in mind is to create a simple microcontroller thingy with a ATmega644. I'm basing it off Admin's $50 robot, and some other tutorials I've seen online. If all goes well, I'll have something that will let me control servos and I/O from sensors relatively cheaply.

[Admin]

wow this topic got way off . . . ummm . . . topic . . .

So my next project in mind is to create a simple microcontroller thingy with a ATmega644. I'm basing it off Admin's $50 robot, and some other tutorials I've seen online. If all goes well, I'll have something that will let me control servos and I/O from sensors relatively cheaply.

hey! so im programming at ATmega644 now actually! we should collaborate somehow . . . the $50 robot is actually based off of what I learned with the 644
but i plan to have some really advanced/expensive features on it so its not the kind of thing beginners should ever attempt . . .