Actions
Forum Thread Details
  • State Suggested Answer
  • Replies 5 replies
  • Answers 3 answers
  • Subscribers 392 subscribers
  • Views 901 views
  • Users 0 members are here
Related

Maze Robot! Heeellppp! :(

the_future

Hi everyone,

trying to build what i thought was a simple maze robot if anyone can help that would be great image

 

aim (simple):

 

calculate and find the quickiest route though the maze

 

questions:

 

* what sensors should be used for mapping and avoiding walls (I suspect ultrasonic or IR for the avioding part)

* what would be the best algorthim and methodology of going about this? (for coding)

  

   ideas in mind so far for methodology

   -> first robot goes through the map for an inital scan of the maze then stores information in memory to then work out the quickest route

   -> first robot scans the maze with a sensor (unsure of sensor) then stores information in memory to then work out the quickest route

 

realised this was a bigger project then i thought haha!

  • 0

    what sensors should be used for mapping and avoiding walls (I suspect ultrasonic or IR for the avioding part)

    Well, I guess this is something you may need to measure first and decide based on tests if you know how your maze is going to going to be. If you don't know, and then you decide IR is your only choice, IR for example won't work well against black walls. Common ultrasonic sensors *usually* work well with a > 2 cms clearance. My take is using different sensors at least two different types that don't have the same weakness and decide based on the info provided by all available sensors and have a way to adjust the decision making -lets say you can disable the IR- and decide with an Ultrasonic and a limit switch.

     

    * what would be the best algorthim and methodology of going about this? (for coding)

    Again, here you may need to decide based on the *available* info about the maze you have. There are plenty to pick from!.

          ideas in mind so far for methodology

       -> first robot goes through the map for an inital scan of the maze then stores information in memory to then work out the quickest route

       -> first robot scans the maze with a sensor (unsure of sensor) then stores information in memory to then work out the quickest route

    Yes this is the right track, if this is about who solves the maze the quickest (after a initial scan). Don't forget rotary encoders to virtually map your maze easier.

     

    I'd suggest also to do a little bit of research on Pololu's blog and forum. Their products focus on robotics and they sometimes host maze solving contests so there is plenty of experience and info for your needs.

     

    Good luck with your project!.

     

    Luis

  • 0

    to add some more context to this I was referencing this guys work for my own maze:

     

    * https://www.youtube.com/watch?v=9qjZIqkusXw

    * https://www.instructables.com/id/Maze-Solver-Robot-Using-Artificial-Intelligence-Wi/

     

    his uses a 2 phase algorithm: 1st phase: Wall Follower Algorithm - Left hand rule and 2nd phase: Shortest Path Algorithm

    however with my maze there is a loop and can get stuck but I have no Idea on how to prepare the robot before it gets stuck.

     

    notes: dsPIC30F4011 being used

  • 0 in reply to luislabmo

    Thanks image could you please expand more on how rotary encoders make it easier to virtually map a maze and maybe how I might go about it!

     

     

      wrote:

    Yes this is the right track, if this is about who solves the maze the quickest (after a initial scan). Don't forget rotary encoders to virtually map your maze easier.

  • 0 in reply to the_future

    Hello,

     

    • Without a feedback measurement there is no way to exactly know where your robot is in the maze because robots need at least 2 motors to go straight/turn. Motors -even if they are same spec- are not going to rotate identically-
    • The robot will have some momentum when moving which adds an -unknown- distance or if hits an obstacle it is going to suddenly stop which has to be accounted for.
    • Lets say your robot hits an obstacle: is it turning 90 degrees and trying again on that direction? 45 degrees? 180? How would you know if your robot rotated exactly what you wanted ?
    • Rotary encoders can be avoided by using stepper motors but these may be overkill for a project like this and also stepper motors can miss steps or stall.

     

    Conclussion: you need a feedback mechanism (rotary encoders usually) that will help you determine with more precision how far your robot has gone in every direction forward/backwards and also to track with more precision rotation. Unless you are making a line following robot (the line will be the one providing the feedback to the IR sensors), I'd advise to add rotary encoders or any sort of feedback.

     

    This video from SparkFun explains on a high level what I'm talking about

     

    Luis

  • 0 in reply to the_future

    the only thing about the video i am learning off is that it is a line following robot and my robot wont be following a line so thanks a lot it makes a lot of sense to include an encoder

    but I feel the algorithms should still work image

     

    my last problem (for now) is that how can I prepare my robot for when my robot approaches certain areas of the maze that make the robot become stuck in a loop ?