Older blog entries for cschur (starting at number 54)

This was the first big test weekend for VacBot-1 up at our cabin in the pines in northern Arizona. It showed me it has still a bit of work to get it to work as planned. First, the robot has bumper only navigation. With only that, its going to get stuck fairly often in a household enviornment. It took me a year of hard work to get PAAMI to be able to run for days in our house without ever getting stuck or trapped. She had not only bumpers, but IR Prox, and sonar and a very advanced 5 layer subsumption architecture. Vacbot has bumpers only, but is an excellent proof of concept for finding out what works and what doesnt!

While I have some more hardware mods, and am in the process of adding the 3 state FSM, here is what I have learned so far:

- when moving from tile floor to carpet, the brush assembly must keep the same height from the surface so it doesnt bog down

- Batteries last a lot longer on tile, nearly 10x longer.

- Bumper only navigation while sufficient for a square competition type arena, is not reliable for a house hold with chairs, furniture, etc.

- a speed of less than 1 foot per sec is best for effectiveness.

- there is a LOT of cat hair its picking up in our carpet!

Write me with your comments:

comets133@yahoo.com

Chris

Hi All,

I think we have it now, we redesigned our Vacbot bumper again, and now it works very well. I think you will find it a design worth using for your own:

http://www.schursastrophotography.com/robotics/vacbot2.html

Thanks for looking!

Chris

comets133@yahoo.com

HI All,

Here is the latest update with lots of new photos of our VacBot project. Take note on our adjustable bumper design we are working on:

http://www.schursastrophotography.com/robotics/vacbot1h.html

The bot is now able to autonomously navigate in the house, but the small rollers just behind the brushes are catching on things occasionally. Left to do is to put on the rear bumper and finish up the FSM in the programming to make it clean in patterns for best results. The FSM will include a spiral state, where it makes the robot move in an increasingly large spiral pattern every five minutes or so until it hits something, then go random again. this should cover the most room in a minimum amount of time.

Chris

comets133@yahoo.com

Hi everyone,

Made more progress on VacBot 1 last night, after a week of shooting the stars, I am now back to robotics in the evening because of the full moon. The bumper on this thing is unlike anything Ive ever built before. Since I am only relying on bumpers for navigation, and not IR or Sonar, They have to be rugged, and dependable. I made a metal frame work, and it hangs with rubber pads on four posts. The idea is by adjusting the bumpers, or using different materials I can get any pressure I want to activate the reed switches. Ill post some imags of this interesting concept - the adjustable bumper - comming soon!

Chris

VacBot progress:

WEll, I decieded to rip out the belt drive, and replace it with motors and a set of nylon gears. I dont think Ill ever try another belt driven system for a robot again! The problem is slippage on the pulleys, causing no end to frustration with precise maneuvers. I have two 230rmp motors left over from the PAAMI project Im going to reduce down and it will give me a precision 1ft/sec travel rate.

I finished the new book "Biorobotics" I got from robot books.com, and have some new insights on insect navigation I can use on the GeoBot project we are starting up on. Did you know that the Calipterus desert and has 20K neurons in its tiny brain? Its eyes have facets of course, but each one has a 2 degree field. Pretty coarse, but thats good enough to navigate. You dont need a megapixel sensor for a robot after all!

Chris

comets133@yahoo.com

Hi All,

More on some progress with VacBot 1, here is a new shot and a movie clip of todays work on the robot. We got rid of the direct drive to the huge 6 inch wheels, and now are trying to use a set of belt and pulleys for a 5 to 1 gear reduction. I dont know what to use for a belt yet, but for now its rubberbands. I got the speed at the half a foot per second I was targeting, and added a back wheel for the sweep height adjust.

Here is the new shot, and the movie at the top of this page:

http://www.schursastrophotography.com/robotics/vacbot1.html

The motors can now be run full speed and have huge torque.

Chris

Hi All,

Finally some new shots of our progress on VacBot. I finished wiring the processor, and motor driver amp up today, and fired it up. Right now, Im just moving it forward, no routines yet. I will need to gear it down a bit for carpets, but it moves fine on the wooden floor. Here is the latest images, the top two on this page and the new movie (with sound!) is the new upload:

http://www.schursastrophotography.com/robotics/vacbot1.html

Getting it all in the box and mounting it was todays hurdle, and getting the gearing and PWM spot on is next!

Chris

comets133@yahoo.com

Hi All,

Ive breadboarded the sharp GPS2 series IR photo detector and tried it out on a homemade encoder wheel printed out with a laser printer. the peak range is very critical, about 2mm and drops off drastically on both sides. Also, even with a 50k pull up on the collector, the range was about 1- 4v out. So a schmidt trigger will probably work. Remember, many PICs have thier inputs on the A port connected in with a schmidt trigger, making interfacing a snap.

Second, Ive given up on the wimpy RIS software that comes with the Mindstorms kit, its a joke. Ive installed the firmware and software for Robolab, and let me tell you, Im about 3/4 the way through the tutorials and it blows the heck out of RIS. It can do simultaneous paths making subsumption architecture a reality.

Third, VACbot is comming along. Ive got all the parts on the PCB and am wiring it in now. I first will have to determine if using PWM I can get the speed of this thing without adding gears to about 1/2 to 1 foot per second. I feel thats a good speed to start. The L298 is a great motor dual H bridge, but why such an awful pin footprint??!

Later,

Chris

comets133@yahoo.com

Hi All,

I am in the process of designing several optical shaft encoder projects for work for a new robotic motor application, and bought some very promising encoder disk photo reflector chips I thought Id pass on to you. The tiny Sharp GP2S series of IR photo reflectors have a range of only a few mm but put out a clean square wave which can be tidied up a bit with a schmidt trigger or compartor with hysteresis if needed. They are inexpensive at 84 cents each and for that price I got a dozen of them! The plan for now is to see how well they read an shaft encoder disk printed out on a laser printer for positional information. Very cool devices!

My vacbot project at home is comming along. Ive got the wheels attached now, and at full voltage its way too fast. But I intend to run it on PWM and so will shoot for a slow half a foot per second for the task of keeping our carpets clean. Im working on a new bumper design that will consist of those 2x3 small circuit boards from radio shack with four small membrane type push switches from Jameco in between on the corners forming a sandwhich. Should be sensitive, yet take up very little space.

Chris

I got around to starting a web page on the vacbot Im working on, based on the "Dirt Devil" motorized sweeper as a platform. This page shows the starting point, before modifications. Ive been busy turning parts on the lathe for the wheel attachements and now the huge 6 inch lawnmower wheels are on it. They have a polyethelene core, and are very light and durable. And for $3 from walmart, they are perfect for the job of carpet roaming. Here is the start page for the new robot:

http://www.schursastrophotography.com/robotics/vacbot1.html

I bought the box and circuit board, connections etc last night, and will be wiring a 16F628 and L298 to control the twin black and decker battery operated drill motors Ive adapted to run it.

Im all ready thinking on how I want the FSM to control the vacuuming action. Put in the center of the room, then it start an overalpping spiral increasing in size until it hits something with its bumpers. then it goes to the random wander mode. THEN every 5 minutes or so, when a timer in the Pic times out, it goes back into the spiral mode. This way huge areas of the rooms will be cleaned quickly. Corners and agains the wall will be the last to be done.

Chris

45 older entries...

X
Share this page