Shop OBEX P1 Docs P2 Docs Learn Events
A perfect Robot — Parallax Forums

A perfect Robot

MilevithiusMilevithius Posts: 4
edited 2008-03-20 18:20 in Robotics

Hello!



I just bought a BOE, and 2 manuals ( ? microcontroller & Advanced sensors) to start my first "touch" into robotics...



The reason why I need your help is·because I have the perfect robot in my mind but don't know if current technology is available to complete it. I have the budget, dedication and commitment to do it.



God kill me if I'm unable to complete it before I die (hehe, excellent line text - loop).



Rule one for my robot:

1. Its aim is survive. All programming will based on surviving.

The characteristics·of my perfect robot·are:

1. Autonomous·surviving capabilities (docking station)

2. Capacity to distinguish between my girlfriend·and me (it would be great if in the morning when my girlfriend leaves it could recognize her and say for ex. "good work").

3. Capacity to·follow orders from speech (voice recognition - any module from parallax?)

3. Capacity·to read (speech to text) my mails/weather/RSS fees with WIFI technology (any module here from parallax?).

4. Some AI. (learn·by errors while moving around the house).

5. Speech phrases learning - example find an algorithm for syntax (simple phrases ex: noun verb noun) by error learning (possibility to go to docking station - ex sensor goes on, so robot can find where docking is placed - if phrase is correct).



I know that some of them already exist, just need to hear a little·motivation from you!

Comments

  • Bruce BatesBruce Bates Posts: 3,045
    edited 2008-03-19 09:58
    Milevithius -

    Here are some thoughts on your proposal. Many more things would need to be considered.

    First, you are not going to accomplish anything like what you've described with a BOE and any microcontroller that I'm aware of and probably none that are available either.

    Second, I would rather doubt that it could be done even with an ordinary PC. With an appropriately sized mini-computer it might be possible

    Third I'd set my budget in the neighborhood of $100,000.00 US and up! That may not include the salary for the software programmer to cover the multi-year effort he/she would have to put into such a platform. This is not a project for someone new to this field.

    Fourth, I know of no software or set of software programs which can do all of the varied things that you want to do. If such programs did exist they would probably be found in academic circles, and thus would be too costly to purchase, or not be for sale. It would probably be part of some research and development work.

    Now, if you wanted to do small parts of what you have in your proposal (see below), that is entirely possible:

    speak simple "stock" words or phrases

    recognize simple sounds and/or some pre-selected words

    connect to WiFi and possibly feed a browser w/ email client

    learn from errors without extensive use of AI

    That should give you some idea how complex and costly such a system might be. BTW, this would NOT be the size of a $100.00 or so kit robot. You're speaking about a rather large platform; large enough for all the requirements you have listed.

    Obviously this it just my opinion, and others may feel it can be done. Sorry if I seemed to rain on your parade.

    Regards,

    Bruce Bates

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Involvement and committment can be best understood by looking at a plate of ham and eggs. The chicken was involved, but the pig was committed. ANON
  • stevenmess2004stevenmess2004 Posts: 1,102
    edited 2008-03-19 11:05
    You should be able to do all of that list except for understanding speech with a propeller chip and a couple of other chips do things like wifi. However, that list requires some fairly advanced programming and also a decent understanding of some maths.

    My advice is to start with something simple and slowly build on it. You will not get everything on the list done in a couple of weeks. You probably won't even get it done in a couple of years unless you are a very good programmer. However, don't let that stop you from playing around and having a fun time. The books you have ordered should teach you a lot.

    And welcome to the wonderful world of parallax productssmile.gif
  • aladasaladas Posts: 18
    edited 2008-03-19 13:55
    Yes - you will die in the attempt, unless maybe someone else gets there first. But you can have a lot of fun and learn a lot trying. The BoEBot will teach you a lot even just taking it up to moving around with whiskers and IR sensors. You will be unable to resist trying the smaller projects which are within your competence, which you will thus stretch. I would suggest you explore the Zoomba (converted Roomba) and the smallest UltraMobile PCs running Windows XP tablet edition (I have a Samsung Q1) as these are small enough to be carried around by a smallish outonomous robot. And there are linux-based PDAs which are even smaller (eg the Zaurus.)

    It seems your tasks for the robot are not far from the things your PC does for you so the environment in which you want it to learn survival is probably your home, where its main perils are having coffee spilt on it or getting stepped on. Let it get the hang of that before you start taking it for walks and dealing with crossing the road. It takes a human child a while to get that far. Are you sure you don't really want a dog?

    All that being said - I think I share a large part of your dream and probably many Stamp users do too. The reality will be humbler. But by the time we grasp this we will be so good at controlling the stamp and its successors that they will be an important component of what finally gets built, though probably not the core of it. A stamp could probably handle the job of controlling a hand - or any other organ, but not all of them.

    Good luck.
  • UghaUgha Posts: 543
    edited 2008-03-19 18:17
    First... I want to make clear that I'm a newbie too. I've only been messing with robotics and electronics for a few months.

    I do want to say though, that I don't have such a bleak view of your goals.

    I think that most, if not all, are absolutely possible... but it would require the power of a desktop PC to work.

    Desktops are too big to strap to a robot... so why not make your robot controlled by a dedicated PC?

    In other words, the robot reads in data, transmits it to the PC, the PC decides what to do and then transmits commands
    back to the robot.

    This is something like I intend to do eventually (and its quite possible... I've already fiddled with wifi some).

    I recommend focusing on the goal of PC communication and control first... then break down your project into various
    systems and complete them one at a time.

    As far as where to start in the world of Wifi control... My setup was with 2 BOEs(one on a boebot) and a $15
    transmitter/receiver set. I'd be glad to give more detail if you wish.

    And to everyone else:
    I know your just trying to explain whats possible and introduce Milevithius to the realities of hobby robotics, but
    you don't have to completely crush his dreams.
  • Bruce BatesBruce Bates Posts: 3,045
    edited 2008-03-19 19:53
    Ugha -

    Just a couple of passing comments, not meant to be a flame -

    Stopping here might have been a good idea: "First... I want to make clear that I'm a newbie too. I've only been messing with robotics and electronics for a few months." Some of us have been around a bit longer than that, and only offer the benefit of our experience. It's been said that experience is the best teacher.

    If your post was at all directed toward what I offered, I also mentioned to Milevithius:

    Obviously this is just my opinion, and others may feel it can be done. Sorry if I seemed to rain on your parade.

    I'm sorry again, but I don't see that as crushing anyone's dreams. Opinions are opinions, just that and nothing more.

    Regards,

    Bruce Bates

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Involvement and committment can be best understood by looking at a plate of ham and eggs. The chicken was involved, but the pig was committed. ANON
  • stevenmess2004stevenmess2004 Posts: 1,102
    edited 2008-03-19 21:25
    Sorry for being negative but Milevithius basically asked if the things were possible. To go through his list in more detail
    1. Autonomous surviving capabilities (docking station)
    This should be reasonably easy to almost impossible depending on how you decide to do it and the environment the robot has to work in. - Easy to Medium

    2. Capacity to distinguish between my girlfriend and me (it would be great if in the morning when my girlfriend leaves it could recognize her and say for ex. "good work").
    How easy this will be depends on finding an easy way to distinguish between you and your girlfriend. - Medium to Hard

    3. Capacity to follow orders from speech (voice recognition - any module from parallax?)
    You can buy chips to do this kind of thing. You simply record several phrase and than it tries to match what it hears against the phrases. I haven't used one so I don't know how accurate they are. - Easy to Medium

    3.1 Capacity to read (speech to text)
    There is an object that does this to some extent on the propeller. I haven't tried it so I don't know how well it works. If it works all right this is Easy, otherwise its Hard because you have to develop it yourself.

    3.2 my mails/weather/RSS fees with WIFI technology (any module here from parallax?).
    You can buy wifi modules (I don't know if Parallax is selling them any more). However, the problem is getting the data from the websites. The ybox project for the propeller has done this successfully though. There are currently IRC bots and clients for the propeller so I would put this as Medium if you know something about how the internet works.

    4. Some AI. (learn by errors while moving around the house).
    This is possible however you will probably need to understand some maths. This is Easy to Hard depending on exactly what you want it to do.

    5. Speech phrases learning - example find an algorithm for syntax (simple phrases ex: noun verb noun) by error learning (possibility to go to docking station - ex sensor goes on, so robot can find where docking is placed - if phrase is correct).
    This may be possible for a very limited set of words. But I think that this one would really have to go back to a PC or something more powerful. Very Hard

    You probably won't fit all of this on a Basic Stamp because you will run out of speed and memory so if you want to do this you will need to move to the propeller or something else at some stage. However, the Basic Stamp is a good place to start as it is well documented. If you do need more processing power than the propeller than you can try some of the single board computers. Some of these have the very latest processors. However, the start needing more power and a bigger bot and the cost will start to spiral.

    So, have fun, learn lots and don't let people like me stop you dreaming but also don't be discouraged when you can't achieve all your dreams.
  • roboman01roboman01 Posts: 5
    edited 2008-03-20 03:04
    I've been tinkering with robots for a few years, and I love the BOE-Bot. It's what really got me into microcontroller-based robotics. Before that, I just played with BEAM.

    I fully support the BOE-Bot as a beginner's robot.
  • DgswanerDgswaner Posts: 795
    edited 2008-03-20 18:20
    I would disagree with you Bruce, you under estimate the power of the dar.... I mean the power of the propeller. Can it do ever thing he wants? no. But with a propeller, or 3 and use of a PC I think every thing he want to do can be done at and ACCEPTABLE level of accuracy.

    The point I think Bruce is DEAD ON, is that this can not be done on a BOE, the BOE is a great bot but with all of the sensors required for this to work there just isn't room.

    That being said, Milevithius I would start with the BOE bot, use it learn from it and when your ready upgrade, I'm on a similar path and I have built and rebuild my bot about 4+ times now, each time getting bigger and more involved. I don't think it would be wise to start with a large bot, they cost alot more and if you make a mistake it's costly.

    If you want the story of my journey let me know I've made all of the mistakes!

    anyway, shoot for the starts! best of luck!

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    "A complex design is the sign of an inferior designer." - Jamie Hyneman, Myth Buster

    DGSwaner
Sign In or Register to comment.