Shop OBEX P1 Docs P2 Docs Learn Events
Science fair projects.... — Parallax Forums

Science fair projects....

ArchiverArchiver Posts: 46,084
edited 2000-05-18 01:25 in General Discussion
Hey all,
I'm required to enter the Science fair next year. I really want to do a
robotics project. I thought about using the question "Can robots be made to
exhibit animal like behavior", then building a number of BEAM, etc. 'bots,
but a friend pointed out that that question was answerable with a "yes" or
"no". Then I thought about asking "what animal like behaviors...". I don't
really know. I'm having trouble thinking of some different behaviors. Do any
of you have any more ideas? I really need to start thinking about it.
-William


____________________________
www.GoRobotics.net
Your premier robotics website.
Robotics books, projects, resources,
links, news, and more!

Comments

  • ArchiverArchiver Posts: 46,084
    edited 2000-05-17 18:42
    Hi William,

    How about seeking light, or even more interesting, power. Many animals are
    photovores and seek light.

    If you had a robot that ran on rechargeable batteries, it could sense low
    power and find a recharge station. To simplify the docking, you could use a
    transformer. The station has AC entering the primary and the robot has the
    secondary plus rectification.

    Now, if you had lots of robots, you could provide a random threshold for
    power to each one. Then they could "mate" -- exchange thresholds and average
    them. Low threshold individuals will eventually run out of power and be
    unable to mate. The remaining population would zero in on the optimum
    threshold (which would depend on lots of factors like number of rechanging
    stations, distance between stations, etc.). This would be easy to simulate
    on a PC as a way to determine the right threshold for your real robot.

    Of course, in a simulation, "mating" could produce a new robot too. So you
    could say if 2 robots meet and they both have sufficient _additional_ power,
    they can produce another robot. Not easy to to in real life though, so I
    still like my originial idea.

    Some animals hide to avoid detection. That would be another interesting
    behavior. If you shine a laser pointer on a robot for 2 seconds, it "dies".
    So it has to hide away from you. If it senses you hit it, it must escape.
    Maybe each hit makes it go slower and after 2 seconds, it is out of the
    game. Cat and mouse.

    Just a few random ideas.

    Regards,

    Al Williams
    AWC
    *Special this week only: Buy one PAK get the 2nd at up to 70% off!
    http://www.al-williams.com/awce



    >
    Original Message
    > From: William Cox [noparse]/noparse]mailto:[url=http://forums.parallaxinc.com/group/basicstamps/post?postID=k8pbxmXykgA2EXwdFi92N6iMqqtJAztOAMAt_Pv2pS93eMunV_CZnuvnIkL2x-0IIYJEtVLFjxuA-n_fQ95ibbGRJQ]william@c...[/url
    > Sent: Wednesday, May 17, 2000 11:54 AM
    > To: srs; Stamps; beam@egroups.com
    > Subject: [noparse][[/noparse]basicstamps] Science fair projects....
    >
    >
    > Hey all,
    > I'm required to enter the Science fair next year. I really want to do a
    > robotics project. I thought about using the question "Can robots
    > be made to
    > exhibit animal like behavior", then building a number of BEAM, etc. 'bots,
    > but a friend pointed out that that question was answerable with a "yes" or
    > "no". Then I thought about asking "what animal like behaviors...". I don't
    > really know. I'm having trouble thinking of some different
    > behaviors. Do any
    > of you have any more ideas? I really need to start thinking about it.
    > -William
    >
    >
    > ____________________________
    > www.GoRobotics.net
    > Your premier robotics website.
    > Robotics books, projects, resources,
    > links, news, and more!
    >
    >
    >
  • ArchiverArchiver Posts: 46,084
    edited 2000-05-17 20:11
    DLC's answer below provides an excellent framework for context-dependent
    environmental cues in robotic-like behavior. The following is an example of
    some
    apparently complex decision-making from the insect world.
    The digger wasp displays apparently "intelligent" decision making. It
    approaches the opening of its burrow, puts down its captured prey at the
    entrance,
    goes in the burrow to check for any intruders, emerges, picks up its captured
    prey
    and re-enters the burrow. Intelligent decision making? No. While the wasp was
    in the burrow, the human experimenter moved the prey 6 inches from the burrow
    entrance. The wasp emerged from the burrow, picked up the prey, again set it
    down
    at the entrance, and again entered the burrow to check it out. This cycle was
    then repeated many times.
    Dennis

    dlc@v... wrote:

    > William,
    >
    > You're having problems because you are thinking like a human and you
    > "think you think like they think". What you need to do is just look at
    > behaviour and build your robot from the bottom up instead of the top down.
    > This is THE topic in behaviour based robotics, how to get behaviour to
    > emerge rather than be programmed. In other words, what is concious behaviour
    > and what is simply observed behaviour that is reaction to the environment?
    > <deleted>

    > For example, a bug's primary objective
    > is wandering about. However, if it sees something in its way, this will have
    > a higher priority to _avoid_ running into it, so this behaviour subsumes the
    > wander behaviour, and so on. You eventually get some kind of behaviour that
    > will emerge from multiple levels of reactions to the robot's environment and
    > the robot's initial goals (go North, or somesuch).
    > <deleted>
    > My advice, think out of the box and lay around watching bees and ants. See
    > if inspiration strikes, I suspect it will!
    >
    > have fun,
    > DLC
    > > Hey all,
    > > I'm required to enter the Science fair next year. I really want to do a
    > > robotics project. I thought about using the question "Can robots be made to
    > > exhibit animal like behavior", then building a number of BEAM, etc. 'bots,
    > > but a friend pointed out that that question was answerable with a "yes" or
    > > "no". Then I thought about asking "what animal like behaviors...". I don't
    > > really know. I'm having trouble thinking of some different behaviors. Do any
    > > of you have any more ideas? I really need to start thinking about it.
  • ArchiverArchiver Posts: 46,084
    edited 2000-05-18 01:25
    William,

    You're having problems because you are thinking like a human and you
    "think you think like they think". What you need to do is just look at
    behaviour and build your robot from the bottom up instead of the top down.
    This is THE topic in behaviour based robotics, how to get behaviour to
    emerge rather than be programmed. In other words, what is concious behaviour
    and what is simply observed behaviour that is reaction to the environment?

    Finally, what is the difference? I will be dissed no doubt for offering a
    book to read, but really, to get a REALLY good handle on low level behaviour
    and robotic behaviour, you should read "Vehicles, Experiments in Synthetic
    Psychology" by Valentino Braitenberg. The book is available for $16 on
    Amazon.com, or may even be in your local library. It is a very easy and
    entertaining book to read and I PROMISE you that you will get great ideas
    from it.

    Several years ago a bright and combative individual by the name of Rod Brooks
    came up with the idea of tying action to perception direction and developed
    the subsumptive architecture for programming robots. In a nutshell, this is
    the process of building behaviours through the interactions between multiple
    priority levels of reactive behaviour. For example, a bug's primary objective
    is wandering about. However, if it sees something in its way, this will have
    a higher priority to _avoid_ running into it, so this behaviour subsumes the
    wander behaviour, and so on. You eventually get some kind of behaviour that
    will emerge from multiple levels of reactions to the robot's environment and
    the robot's initial goals (go North, or somesuch).

    I wrote a paper for Parallax on subsumptive programming using Finite State
    Machines on the Stamp computer that is available on their site under the
    robotics section. It (hopefully) does more to explain these ideas and show
    how to program a robot using them, specifically, the Stamp II BoE-Bot.

    My advice, think out of the box and lay around watching bees and ants. See
    if inspiration strikes, I suspect it will!

    have fun,
    DLC
    > Hey all,
    > I'm required to enter the Science fair next year. I really want to do a
    > robotics project. I thought about using the question "Can robots be made to
    > exhibit animal like behavior", then building a number of BEAM, etc. 'bots,
    > but a friend pointed out that that question was answerable with a "yes" or
    > "no". Then I thought about asking "what animal like behaviors...". I don't
    > really know. I'm having trouble thinking of some different behaviors. Do any
    > of you have any more ideas? I really need to start thinking about it.
Sign In or Register to comment.