Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!think.com!snorkelwacker.mit.edu!stanford.edu!csli!cphoenix From: cphoenix@csli.Stanford.EDU (Chris Phoenix) Newsgroups: comp.robotics Subject: Re: CALM (computer-aided lawn mowing) Keywords: feasibility, beer Message-ID: <20073@csli.Stanford.EDU> Date: 21 Jun 91 01:54:57 GMT References: <1991Jun20.192334.24623@auto-trol.com> Distribution: na Organization: Center for the Study of Language and Information, Stanford U. Lines: 61 In <1991Jun20.192334.24623@auto-trol.com> bridon@auto-trol.com (Brian Donahue) writes: >i have this (maybe ridiculous) idea for a 'automatic' lawn mower. >The idea would be to mount a mulching lawn mower on a 'computer-guided' >platform of sorts. ... use an NC-type algorithm to guide your mower and Maybe I'm just brain-dead today... what's an NC-type algorithm? >Being a robotics neophyte, i have no idea how to approach this problem >or what complications may make the problem harder than it sounds. >For example, would i need some sort of tracking system? or can i assume >that for every turn of the wheel the lawn mower goes a certain distance? >(how accurate can i get w/ no tracking system?) You definitely need some kind of tracking system. Consider: 1) Wet grass is slippery. 2) The yard is bumpy, so the wheels actually travel slightly different distances. I doubt anyone could build an unguided system that would hit a 1-meter target across a 10-meter yard, without some massive sensing/feedback. Maybe a walking-type system would work, but I don't think it can be done with wheels. (If I'm wrong, someone please tell me, but we had problems making a synchronous-drive base roll straight on carpet, and this seems like a much harder problem.) >or is this totally ridiculous & not worth pursuing? It's not totally ridiculous. It'll be a great learning experience. :-) Seriously, it depends on how much time you want to spend on it, and how general you want the solution to be. My first inclination was to advise you to build some sort of position-and-orientation sensor, that could tell the platform where it was well enough to use a general path-planning algorithm. I've been in school too long. So, I'd advise building a really simple sensing system, consisting of reflectors all around the yard and a machine that only knows enough to head for a reflector and turn when the PC tells it to. Mark off the yard with a grid, making each "strip" (crossing strips form the grid) the width of the mower's blade. Where each strip intersects the edge of the yard, put a reflector. If there is an obstacle in the middle of a strip that obscures the strip's reflector, put a reflector on each side of the obstacle as well. Now sit down with graph paper and draw your yard, and then draw a path that covers each square of graph paper without an obstacle in it. Put hardware on your mower to do two things: 1) Move toward a reflector that is in front of it. 2) Turn 90 degrees. 3) Sense when it rolls past a reflector to the side of it. Program your PC to count the reflectors it passes and turn at the appropriate points to follow your path. Another possibility: Bury a wire under the yard, and make your machine follow the wire. This is simplest--you don't even need a computer. But it takes a lot of wire for a big yard, and you have to worry about something breaking the wire. (Of course, you have to worry about something removing the reflector, too.) I don't know what kind of circuitry you need to make this work--I would guess an RF generator you hook up to the wire, two antennas on the machine, compare the signal strength from the antennas to decide where to turn. If anyone has any feedback on anything I've said, please tell me, either in e-mail or on the net. Thanks! -- Chris Phoenix cphoenix@csli.stanford.edu #insert #insert #insert bomb detonate assasinate Libya Bush nuclear PCP <- Just to waste the CIA's time