Shop OBEX P1 Docs P2 Docs Learn Events
Obstacle and surroundings mapping. — Parallax Forums

Obstacle and surroundings mapping.

BotdocterBotdocter Posts: 271
edited 2010-03-18 23:50 in Propeller 1
I want to give my balancebot a sort of memory for obstacles.
Something like Ping)))Dar.

Only this program should be able to let the bot navigate from memory. Obstacles that it's sensed before. Could i use a mouse for coordination (or 2)?

Is there allready something like that?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
1 Parallax Propeller Robot Control Board
1 Memsic MX2125 accelerometer/ tilt
1 Parallax Ping))) ultrasonic sensor

a few motors and a whole lot of chaos!

Comments

  • SRLMSRLM Posts: 5,045
    edited 2010-03-17 03:52
    Nope, you'll have to make it yourself. You'll almost certainly want to get some encoders of some sort.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Powered by enthusiasm
  • BotdocterBotdocter Posts: 271
    edited 2010-03-17 04:58
    Yes i know. I have ir sensors to use for the wheels but would rather use one or 2 mouses would it be posible to navigate at all with a mouse and is it possible to connect 2 mice simultaniesly?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    1 Parallax Propeller Robot Control Board
    1 Memsic MX2125 accelerometer/ tilt
    1 Parallax Ping))) ultrasonic sensor

    a few motors and a whole lot of chaos!
  • rjo_rjo_ Posts: 1,825
    edited 2010-03-17 06:18
    If want to you wait a about a century... I'm working on it... using a Tsl1401. If I get anything to work, I'll be posting it. If you get anything to work... please post it.

    I was reading ... somewhere... that there is a general problem having to do with sensor integration and control theory that creates fault conditions, regardless of how you integrate accelerometers and gyros. So, you are looking at a potential problem set that hasn't been completely solved!!!! Good things have been done... and plenty of successful applications have been create for a very long time, but there appears to be room for fresh thinking.

    We use vision as a feedback for the body's gyros and accelerometers. Since blind people do get around pretty well, it doesn't appear to be completely necessary... but highly desireable.

    When you are choosing sensors... you might want to consider the tsl1401.

    On to the future!

    Rich
  • BotdocterBotdocter Posts: 271
    edited 2010-03-17 18:58
    Indeed! I had a look at it before. That's the linescan sensor that is used in the sensor from parallax.
    But you probably need a lot of extra hardware to run it and then the ready made sensor will probably be cheaper.

    I have also been looking into the mouse cam thing. Adns2610 chip
    with a wide angle lens you can use it as a camera. Also the 2620 has allready been hacked for the propeller. In theory iy could be used for glyph tracking or blob tracking.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    1 Parallax Propeller Robot Control Board
    1 Memsic MX2125 accelerometer/ tilt
    1 Parallax Ping))) ultrasonic sensor

    a few motors and a whole lot of chaos!
  • BotdocterBotdocter Posts: 271
    edited 2010-03-17 21:00
    rjo_ said...
    If want to you wait a about a century... I'm working on it... using a Tsl1401. If I get anything to work, I'll be posting it. If you get anything to work... please post it.

    I was reading ... somewhere... that there is a general problem having to do with sensor integration and control theory that creates fault conditions, regardless of how you integrate accelerometers and gyros. So, you are looking at a potential problem set that hasn't been completely solved!!!! Good things have been done... and plenty of successful applications have been create for a very long time, but there appears to be room for fresh thinking.

    We use vision as a feedback for the body's gyros and accelerometers. Since blind people do get around pretty well, it doesn't appear to be completely necessary... but highly desireable.

    When you are choosing sensors... you might want to consider the tsl1401.

    On to the future!

    Rich

    well pingdar is able to do it but just doesn't have the memory and obstacle avoiding.
    Or isn't that what you mean with the sensor problem?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    1 Parallax Propeller Robot Control Board
    1 Memsic MX2125 accelerometer/ tilt
    1 Parallax Ping))) ultrasonic sensor

    a few motors and a whole lot of chaos!
  • rjo_rjo_ Posts: 1,825
    edited 2010-03-18 02:25
    You are right... and I wouldn't send a bot anywhere without pingers. . There is no such thing as a perfect sensor. If you have two different sensors doing basically the same thing... then when one is in a fault condition... the other can help identify that condition. Data rate and angular resolutions are important and they are different. In theory, our job is to exploit those differences to our advantage... not worry about which is better or worse but take advantage of the basic differences between them in carrying out our particular tasks.

    When I get a little farther along I will be able to be more concrete and useful on this point. For now, accept the idea that even in robotic vision applications... the robot will occasionally have double vision and it will occasionally fail catestrophically in environments that are not suited to the algorithm being use to drive the vision system. The same is true for almost every sensor out there... but in sllightly different ways.
  • DgswanerDgswaner Posts: 795
    edited 2010-03-18 22:42
    I have been wanting something like this for a long time, I understand how I want it to work, but I lack the time and skill (time to learn) it takes to make make this happen. I started a huge thread a couple years back that resulted in a lot of talk but nothing came of it.

    I agree that you need to use multiple sensors to detect objects, if I rely on just Ping sensors my couches are invisible to my robot.

    Idea: use Odometry to keep track of the robots position. have a built in list of coordinates that are do not enter zones. Stairs, the TV, etc.

    This would allow a bot to travel blind if the area was never used, but that's not going to happen and that's where the sensors come in, allowing it to avoid the obstacles. my idea for compensating for sensor errors and "ghost" is, when a coordinate gets flagged as an obstacle it gets assigned a value of say 5, with each cycle that sees that coordinate as an obstacle the value is reset to 5, each time the coordinate is clear, the value gets decremented until it reaches 0, and coordinate with a 0 value is free for travel. This would prevent a person walking in front of the bot or other non permanent item getting mapped permanently.

    to compensate for the odometry error, you could have set positions where the bot checks it location against a known location. ie: going down a hall, it measures it distance from an edge or the end of the hall and corrects it position.

    Thats the direction I want to head, if I become fluent in Spin over night and some how get this working I'd be glad to share!

    good luck

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    DGSwaner
    "When in doubt, use C4" - Jamie Hyneman, Myth Buster
  • BotdocterBotdocter Posts: 271
    edited 2010-03-18 23:50
    My problem is that i don't have the spin programming skills yet.
    I'm learning as i go.

    I was thinking about putting a room map on sd card and then it will avoid the furniture.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    1 Parallax Propeller Robot Control Board
    1 Memsic MX2125 accelerometer/ tilt
    1 Parallax Ping))) ultrasonic sensor

    a few motors and a whole lot of chaos!
Sign In or Register to comment.