Australia
Asked — Edited

Using ARC To Naviagate Robot From A To B And Return

Sorry about TYPO - I mean Navigate...

I'm trying to get the Robot to go from the Lounge room corner to the kitchen - stop for a time period then go back avoiding obstacles along the way. I'm after a way to achieve this using ARC.

One idea I had was to use the recorder but that doesn't work - too much data. Another way is to script him along whilst using the ping obstacle detector (not sure you can do that). Anyway, I'm really keen to achieve this. If there is a cloud example or a series of EZ-B script commands I would really appreciate the example.

Thanks ;)


ARC Pro

Upgrade to ARC Pro

Get access to the latest features and updates before they're released. You'll have everything that's needed to unleash your robot's potential!

PRO
Synthiam
#1  

You'd require a beacon and gps. Gps isnt entirely accurate enough indoors. So a beacon setup would be most reliable

Spain
#2  

How about a combination of time intervals and orders with the compass direction that connects the port i2c, I have seen many useful commands like "compass forward" or "compass turn" I think without the compass the robot is doomed to start from a position and heading and simpre exact same but with copass only need a boot right place no matter the direction.

Australia
#3  

Thanks guys, Both look worth a try, I'll need to buy a compass R2D2 and then come back to you - DJ, how does a beacon and GPS setup work ? Is there something I need to purchase ? eyeroll

Australia
#4  

DJ - can you explain more about the Beacon set up please ?:)

Australia
#5  

I just read a thread about an IR sensor homing in on a beacon - I see GlickClick was also asking for help on this subject. Does anyone know where to can get this setup ?

:)

Spain
#6  

Not if it relates to an infrared beacon but I think those things work better in a straight line without obstacles, could also decorate your house with glyphs on the walls and recognize rooms or kitchen, but personally I like the idea of ??a robot self that needs no more accessories or adapted scenarios, it better be the robot who suits my house. Would you describe as the tour of the hall to the kitchen or draw a picture? I wonder if it is a journey with many twists or rather direct. It would also be interesting to measure the speed of your robot in seconds, for example: 1m/segundo. Knowing this and timing can make a tour script with sequences of walking straight time sequence, spin, time sequence, [1m is 1000 milliseconds, sleep (1000)]

Spain
#7  

for example this route:

User-inserted image

This way of planning the route also requires the measurement of time in milliseconds it takes for your robot to turn 90 or 45 to draw diagonal lines, however this system is not very accurate because they could skate wheel and deviating from the start and trigger a route increasingly erroneous. So I think that the compass could act correct, if not to go in the right direction. And what happens when an object has been moved and hinders the programmed route? Person or is on route? We had the route programming combined with compass, but now we need the actuaccion distance sensors and variables that allow the robot avoid obstacles and return to their path, calculating how far it has to go back to it. I think the best navigation system is the combination of multiple systems.:D

New Zealand
#8  

What about having him roam freely in the general direction... until he finds a glyph on the wall of the room he has been sent to find?

Australia
#9  

Thanks Guys ..

I wonder how Shinsel does it with Loki ? The beacon idea looks the most promising - as DJ said - but I'm looking for all kinds of answers to this challenge.

The route Bob has to go is pretty much an upside down L tilted 45 deg so I could have him travel in a straight line for 5 meters (x amount of milliseconds) then turn left (measured degree of turn by milliseconds) travel a meter distance and stop. The glyph idea is sticky because it will be dim light in the morning and the camera that came with the kit is average (like a baby monitor camera). I've messed around quite a bit with all the video stuff and it is demanding on light and distance balance. I could try the camera on the netbook - but that's probably not much better. This is all new territory and is going to take me looking at plenty of things that have been tried. Multiple systems looks a headache - but may be the only option. Thanks for the tips and things to try.

If I find something straight forward and up to the job you guys will be the first to know.:)

Australia
#11  

Also found this... Forget beacons and gps. With this technology..Robots can not only avoid obstacles around your house but can learn and avoid people that might walk in its way. I have no idea how it works but it sure is going to be fun finding out how. You need to use Kinect software. One catch - the kinect hardware is around $200.

Kinect Navigation

Australia
#12  

OK everyone, a MS Kinect sensor is on my xmas list - you can pickemupanywhere for $92 here in Australia. If anyone can do better please let me know.

Cheers

:) :) :)

Spain
#13  

I think it's a good price for kinect and incehible serious progress, but first we need to ez-b supports kinect, I have read something about future plans between ez-b and kinect but can not find the thread.

United Kingdom
#14  

This is interesting someone needs to write a script

United Kingdom
#15  

@DJ here is an idea if you could draw on the floor map some basic shapes to match your robots surrounding then when the virtual bot gets close to or touches a boundary shape on the map it can change the direction of the real bot to avoid it or follow a set path round the shape and go to a final destination.

I would still want to use a ping sensor to avoid any unforseen contacts obviously and maybe have some points on the map corresponding to real world points where things can get re-calibrated. I've added an example

Maybe just have shaded no go areas

User-inserted image

Australia
#16  

Hi winstn60, I had a similar idea to solve getting Bob from Lounge to kitchen then after testing discovered a major snag. There's a problem with floor map..unless I missed an update, Floor Map is useless for this kind of project as it will not save after only travelling a few feet. Something about running out of memory which I find extrordinary - however, thats why my lates project is on hold.

Cheers:)

United Kingdom
#17  

Thanks for the headsup Hazbot

United Kingdom
#18  

HI Hazbot

Just got back from my holidays and been catching up on the forum .

Chechout this robot and nav program it may help you .

www.pioneernet.net/johnc/cybertcoding.htm

Its very informative hope it helps.

Steve_C

PS your bots been eating his weatabix m8 woow hes grown :)

Spain
#19  

If we could make a map of the site using a ping sensor with a servo mounted for rotation would be great. I think we would need rotary encoders on the wheels to provide reliable data of the robot position. Also we need a calibration conrol to coincide with the ping sensor in angle showing the software and the readings were correct. I suppose it would be a great challenge level programming and mathematical graph above convert sensor measurement data to a simple drawing table. But it would be a fantastic breakthrough in robot navigation. The idea is that the robot walk forward while the ping sensor moves left and right taking distance readings.

:) :)

United Kingdom
#20  

HI R2

Please see above link . that guy has already done it . we have just got to convert it to EZ-B . and bingo there is the mapping nav program we all need.

Steve_C

United Kingdom
#21  

Wow thats my idea bought to life as you say Steve_C we just have to get it to work good spot

United Kingdom
#22  

@winstn60

Yes like r2 said we need encoders so we can rotate 90 degress and scan in front and save the result to the map array . All the programing is there, just need to intergrate the sdk.

Steve_C

PS Have you checkedout the other features of his bot . some good stuff there.

Australia
#23  

Brilliant find Steve.

I love the idea of the robot using ping to calculate his movement forward and following a map. What I like about using ping is your not relying on light as you would with a camera, the robot could move around in the dark. If one of us could come up with the EZ-B adaption of this thing it would make our Robots totally awesome and my dream to have Bob come over to the kitchen for my morning weather and traffic reports when I get up for work could come true.

Lets all get behind this and see if we can come up with a working template we can all use. Thanks everyone for persisting on this one - we are almost there . :D :D :D

United Kingdom
#24  

@Hazbot

If you use the mood routine as well he may tell you to "get your own traffic report" . and go back to sleep:)

#25  

another way to navigate is roborealm is has a navigation software build in to it and Steven from RR says its amost done interface to EZB,plus also there is fudicals can be used from roboream and web camera that EZB has

Spain
#26  

I love the idea of ??sailing using the resources provided to us ez-b, I mean everyone, because with ez-b can do great things with a small investment of time, and what's better: our imagination and ez-b can create any purpose. No need to be a programmer or engineer to master ez-b, so I think the combination of compass, scripts, encoders on the wheels, and one or more ultrasonic sensors, once made the optimal combination can give very good results. Personally I would try to raise the compass above the robot with a telescopic system nonmetallic, to get her away from magnetic fields or servo motors, as long as you find a way to alrgar cables i2c port connection, commented some ferrite cores that such a shielded cable or coaxial?:) :)

Australia
#27  

Thanks for joining the journey on this R2D2

Presently I'm not happy with the compass idea because I've got magnetic fields from speakers all over the place. Even using a real compass in my lounge room is unreliable - I've tried it. I still think mapping using ping is the go but I'm still sorting through that one. It's too bad Floor Map and Ping couldn't work together some how. As far as I know Floor Map still has the problem recording distance travel. I was devastated about that.. it seemed the perfect tool.

#28  

they make a post to raise the compass higher just a long threaded rod and a plastic plate,i i think it be cheaper to make one from parts should not need shielded cables on the compass on a robot ,i have it 6 inches away using a pic processor and normal wires can post the photo of it should be able to run long wires on I2C buss,so 2 problems it may have one is the pull up resistors on the 2 lines or a cheap compass having a week signal

Australia
#29  

Steve C - I downloaded the navigation script from the link and running it through VB-6. It's tidy but will take time to figure out..;) Working with VB is a whole new ball game for me... but my next project using the EZ-B is something I'm not giving up on.

I also found this and listed the site below. If somehow DJ (who's up with programming code and VB) could incorporate this into the ARC, it would be the biggest breakthrough in development for EZ ROBOTS for 2012.

Quote " Once the robot has provided sufficient data about its surroundings, the user can use this map to send the robot to any area of the environment and the robot will avoid all obstacles on its course. This is done by clicking on the goal position of the robot with the right mouse button (draws the green dot in Figure 4) and pressing the "Determine Path" button in the VB interface."

Map navigation