Shop OBEX P1 Docs P2 Docs Learn Events
Pseudo-Intelligent Autonavigation with BoeBot — Parallax Forums

Pseudo-Intelligent Autonavigation with BoeBot

markwardmarkward Posts: 7
edited 2008-01-24 21:50 in Robotics
First project:
Standard Boe Bot, HM55B compass, PING))), eb500 BlueTooth, ir bumpers, ServoPAL. Unit navigates w/standard ir bumpers as primary avoidance mechanism. Ir bumpers are queued 8 slots deep to determine ambiguous obstacle state whereby a 180 degree ping sweep is performed. Data is fed via eb500 to StampPlot for view. Maximum ping distance is translated into a relative bearing and direction and then into an actual heading value to which the boebot is directed to turn. Current problems: 45 degrees to smooth surface is sensed as max distance (seen in video). Possible enhancements include the addition of directional biasing code to direct overall roaming performance.

Video at:

Comments

  • Larry SutherlandLarry Sutherland Posts: 77
    edited 2008-01-20 21:07
    Nice job!
  • WhitWhit Posts: 4,191
    edited 2008-01-21 16:44
    Very nice.

    At 45 degrees, a 0 degree reading should also be max (and 135 degrees and 180 should be max also on smooth surfaces) could you - if 45 and 0 equal max or 135 and 180 equal max. rotate the bot 22.5 and read again?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Whit+


    "We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
  • markwardmarkward Posts: 7
    edited 2008-01-22 03:42
    Absolutely, but the problem is that the 45 degree (and 135 degree) angle to the surface point can come at any point in the scan. the key is to process the data around the seemingly anomalous reading. By this I mean that you'd show a nice solid surface for a few readings then it will just shoot off into space. My code is simplistic in that it only selects the furthest reading as the suggested heading. This seems to work out in the end though due to the ir bumpers. They do a great job of repositioning for a better scan. As a side note I noted that when you scan directly into a corner, the signal will bounce around a bit and appear to "square-off" the corner when you plot the readings in XY coordinates. Pretty cool actually. Also noted that the ir bumpers do a lousy job on my black university chair... This is great stuff, really enjoying it. I've moved on to a balancing bot using ping)) as seen in a few other posts. The servoPALS work like a champ also. Thanks for the suggestion Whit!
  • shimniokshimniok Posts: 177
    edited 2008-01-23 02:13
    Very cool! Would love to know/hear more about this one...

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    ---
    Michael Shimniok

    Bot Thoughts robotics chronicles
  • DgswanerDgswaner Posts: 795
    edited 2008-01-24 21:50
    Markward, your bot is behaving exactly like mine, different configuration (i'm only using a PING) but my bot will avoid most items..... plush couches are hard for my bot, and when It's heading for a wall at a 45 it won't see it until it's hitting it. my way to compensate for this was to go through a loop and if the bot is going straight for X amount of counts it does a sweep with the PING even tho nothing was detected straight in front of it. this would allow the PING to sweep to an angle greater than 45 and see a wall.... if there is one. this does cause the bot to possibly turn, but I find this acceptable. I think that IR and PING combo is a must. even then you can have "invisible" objects.

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

    DGSwaner
Sign In or Register to comment.