hi Crosh it was interesting to know your ideas......To my knowledge the robot will be moving on an even ,plain surface.
Ok - when I read what you were doing, I kinda figured that - but I had to ask...
i am thinking of putting a motor for each wheel because that would also help to have more power and speed .presently i have planned to use encoders on ach wheel.. i should achieve a speed of 1m/s. Yeah i understand balancing wil be an issue but i will try to make a ansys model before going for the practical one.....
Balance will be an issue, and coordinating the motors will be an issue - depending on what you use for wheels and steering, of course.
I had to ask you Mr.Crosh...how far is a gantry crane design better compared to a wheeled approach ..if i m not wrong ,by a gantry crane do u mean this
Yeah - something like that; it wouldn't need the whole wheels and frame thing if you could mount the rails overhead for the rest of the device, but for the span it would have to be fairly big. Of course, you could instead make the robot be a small sized "gantry crane" style machine - in other words so that the middle is open so you can pick up and move the pipe; in this manner the machine would only have to be somewhat larger than the pipe you are moving; a frame 1.25 meters on a side would be OK I think - with wheels at the four corners. If the wheels could be rotated in sync with each other 90 degrees, and all four wheels driven - you could move in a cartesian grid fashion. Suspended from the top-middle of the frame could be a "gripper" or something that could reach down and pick up the pipe (maybe with an electromagnet, or some other kind of device on the pipe to allow each attach/detach from the robot, while not interfering with your testing system).
If you had cameras or other optical sensors pointing downward at the middle points of the square at the base of the robot (one or two sides would need to be hinged, of course, to allow for the pipe to enter in the "cage" of the frame), then you could paint or otherwise afix a grid to the floor of the surface, and count line crossings or do other image processing to position the robot.
there is a budget constraint and installing a whole setup of crane for 20m X 20m will need a huge budget...I am not much experienced wit robo vehicles so can tell me that is it not possible to achieve accuracy of atleast 5 to 10 mm in the vehicle movement ,for the wheeled approach???
Ok - well, I didn't know the budget, and you're right that a large gantry crane would be very expensive (and likely overkill). But a wheeled frame robot, of a gantry crane style design (maybe made from t-slot?) wouldn't be too expensive, if you kept it around 1.5 meters on a side. As far as accuracy is concerned, I am not sure you will be able to get sub-centimeter resolution or not; it would depend on a number of factors (surface, wheels, wheel slip, motors, gearboxes, etc). I think you could get at least centimeter resolution. Possibly with good design you could get below that (understand that I am looking at this as a hobbyist - so if you have access to commercial level engineering and design, my hesitation might be unfounded).
For further tracking of exact position of vehicle ,,i am also planning to use a laser sensor which can be fixed on the test bench(of ultra sensor )on top of a small turn table(to align in towards the vehicle) which can be used to track the position of robot...although i do not know whether this would be good idea..
I think it would definitely be something worthwhile to pursue; probably a combination of tracking approaches (external and on-board) will get you the resolution desired...
Also can you please tell me whether u mean installing cameras on each corner of 20X20m area
I think you may have misunderstood what I was getting at; what I was meaning was putting a camera on the robot, and have it point down on the robot's base to find the QR code which would be painted on the floor the robot is traveling on. The code would act in a manner of a "fiduciary marker" (
https://en.wikipedia.org/wiki/Fiduciary_marker), to allow you to know where and how to place the pipe. You could perhaps put a camera facing downward at each corner "leg" of the 1.5m gantry-frame robot device I wrote about earlier; these cameras would be positioned to look down at the floor, and the QR codes would be stickers or something pasted to the floor. The gantry could move along the grid, then when it found the QR codes (which would be coded with position and possibly orientation information), it would know that it was close to the center-point, and then use some other system to gradually move the robot to exact placement before setting down the pipe or picking one up.
Another note on the gantry design - something to help you envision it better (perhaps): Think of the machine as a four-legged table. The table is 1.5 meters in width, depth, and height. The "top" of the table is mostly open (think of this frame as being made of t-slot, or welded alluminum box extrusion, or similar). Motorized wheels are attached to the bottoms of the legs. They may have some kind of steering system, or they may be the "omni-wheels" as you originally noted. The cameras (for the QR codes and grid sensing) are also on these legs pointing down. You might have at the top corners (between the legs and the edges at the top) corner "bracing" to give it a bit more strength. In the center of the top, attached via a cross-brace or such - would be a vertical lifting and gripping system for the pipes).
and with help of QR code iproject a symbol at the exact co-ordinate and again a robot with a scanner or some optical device to search for the Symbol..Or is it possible that using a single Camera on the vehicle i can make a map of the area and then search the co-ordinate i need???????
Possibly. There's a system out there (patented and protected up the wazzoo) that was developed that used a very small camera mounted in a pen-like device to write on special paper that had a micro-printed pattern of dots printed on it. What it allowed you to do was write on the paper, and the "pen" knew where it was on the page, what page it was one, etc - and it had a memory to store this information. It was designed for note-taking, and Leap Frog (the kid's learning toy company -
https://shop.leapfrog.com/leapfrog/) made a version of the pen (called the FLY/FLY Fusion) that allowed you to play games and other learning tasks (other companies made more serious versions for businesses and students for note-taking, etc).
Now - the secret of all of this was the paper, and the printing system (they had a piece of software that allowed you to print your own pages as well) - more can be found here:
**broken link removed**)
https://en.wikipedia.org/wiki/Anoto
The dot pattern that was read was unique based on the page, the company who made the page, etc - and apparently the amount of surface area all those pages could cover was VAST (millions of kilometers in area - something like that!); so these special pens (and even other optical readers, I imagine) could not only tell what page they were on, but who made it, where it was on the page - and if it had the data stored, what was written there.
All this is to say that - yeah; you could, in theory, put some kind of a special dot pattern or such on the area of your floor, and use a single camera on the robot to look at this dot pattern and tell where it was located at. I am not sure how you would do it (maybe a pseudorandom number generator could work to generate the dot pattern, and then a fast search algorithm to search the "string" from the camera and find where it aligns on the dot pattern?); I would suspect that any method might run into the Anoto patents and IP, too (so unless this is a one-off device that you don't plan on selling - don't do it - but your choice, ultimately).
I hope this gives you some further ideas on how you can accomplish your goal - I think it is possible, but you'll have to give it some more thought and consideration. Good luck.