 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : Poor mans RTS
Page 1 of 2   |
|||||
| Author | Message | ||||
Gizmo Admin Group  Joined: 05/06/2004 Location: AustraliaPosts: 5012 |
Hiya In the process of building a autonomous lawnmower. I hate mowing, I need to mow for a few hours a week on the ride on and my back is not up to it anymore. The mower is all electric, and has a cut width of about 1000mm. I'll be using Ardupilot to map a cutting path with waypoints to cut rows about 500mm to 700mm wide. Looking into positioning technology. From what I can gather, GPS is good for about 3 meters accuracy, no where near accurate enough for a lawn mower. RTS GPS with a base station is good for around 10mm to 25mm accuracy, way more than I need. RTS is expensive, and requires two units, one for the base station and the other is the rover. As I understand it, RTS uses two GPS stations, and these not only get the time information from the satellites, but also measure the carrier wave phase angle to give a higher precision. The base station is set up at a known GPS location, measures the satellite given position, and sends the error/offset in location data to the rover, which uses the same error data to offset its own GPS signal to give a final result. Could this be done with two plain old GPS modules to give a accuracy under 50cm? One module as the base station, transmitting the offset info to the rover. Just looking for feedback, I'm new to this and could be going down the wrong path. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
led-bloon Senior Member  Joined: 21/12/2014 Location: AustraliaPosts: 202 |
So long as you are going down the wrong path, not your mower?  Can't answer technically on GPS, but thinking more of control via local (on your block) signals/markers, but still can't come up with any suggestions. Maybe someone else can. led Miss you George |
||||
phil99 Guru  Joined: 11/02/2018 Location: AustraliaPosts: 1773 |
"Could this be done with two plain old GPS modules to give a accuracy under 50cm? One module as the base station, transmitting the offset info to the rover." That sounds plausible. Observing my GPS the altitude can go up and down by a few tens of meters but the Lat. and Long. drift by just a couple of meters. At the start of mowing the base records its current Lat. and Long. thereafter the the base transmits the differences from that and the rover adds them to its GPS data. . |
||||
Revlac Guru Joined: 31/12/2016 Location: AustraliaPosts: 961 |
The GPS systems used for farming appear to be quite a bit more accurate than I thought. Might get some ideas from those. https://www.fwi.co.uk/machinery/6-budget-gps-guidance-system-options-for-farmers Cheers Aaron Off The Grid |
||||
| greybeard Senior Member Joined: 04/01/2010 Location: AustraliaPosts: 157 |
ignore this, I just suggested what had already been suggested using two gps modules Edited 2022-10-29 19:12 by greybeard |
||||
TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 5882 |
Provided that both modules are identical and both have a very similar view of the sky, it should be good. It is something I have considered for mapping my block but in that case I am in the driving seat. JIm VK7JH MMedit  MMBasic Help |
||||
| bigfix Senior Member Joined: 20/02/2014 Location: AustriaPosts: 124 |
There is a project with a DGPS lawnmower: Ardumower DGPS Modules are not really cheap... RTK2B modules |
||||
| Gizmo Admin Group Joined: 05/06/2004 Location: AustraliaPosts: 5012 |
Picked up a couple GPS modules, same model, from Jaycar today. I'll set these up on a couple ESP32's, one as a base station, other as a rover. The ESP32 has built in wifi, so that makes communication between them easier. Then I'll do some testing. I guess the main test will be to see if both are stationary, they have the same offset at the same time. If that passes, I'll get the rover to request the offset form the base station, apply that to its own readings, and see how much it appears to be wandering from a stationary position. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 3496 |
I will follow this, seems interesting. Automatic lawnmowers I know work with some form of electric fence, dug into the perimeter of the lawn. Maybe an alternative could be in using mirrors at edges detectable with a scanning laser beam on the mower, similar to what land meters do. PicomiteVGA PETSCII ROBOTS |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 5882 |
Those fence loops are one of the greatest sources of radio interference, worse than normal electric fences, and the curse of radio users everywhere. I want to know how it handles the dog asleep in the sun... Jim VK7JH MMedit  MMBasic Help |
||||
| phil99 Guru Joined: 11/02/2018 Location: AustraliaPosts: 1773 |
"I want to know how it handles the dog asleep in the sun..." A very important point. Ultrasonic range finder? Parking radar from a car? Flexible bumper bar with microswitches? (flex distance needs to be greater than stopping distance) Lidar is probably over the top. Then report back (via the GPS base as it already has a link to that) that all is not well. Edited 2022-10-31 07:27 by phil99 |
||||
| Gizmo Admin Group Joined: 05/06/2004 Location: AustraliaPosts: 5012 |
Those things basically drive in random directions. This mower will use waypoints to mow in rows across the yard ( paddock ). All the waypoints are set up before it mows, so it knows about any fixed obstacles. It will have a bumper with switch, so if it bumps into anything unexpected, it will stop and send me a notification. To mow in nice rows I'll need a compass and reasonable accurate GPS. At a waypoint, the rover will turn to face the next waypoint using the compass, the head in that direction, checking its on a line from the old way point to the new way point using the GPS and some trigonometry. I talk about the hardware side of things here. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
bigmik Guru Joined: 20/06/2011 Location: AustraliaPosts: 2867 |
Hi Gizmo, I am no expert on this but why not use standard GPS positioning, if it strays by half a metre this time it will most likely catch up on the missed section at the next mowing session. Or you can manually drive Over that section if it offends your senses (or the missus) Out of interest what size are you expecting to mow? I ‘think’ GPS is better on non overcast days so you could pick your days at times. The way I see it if the automaton does 90% of the job you can enjoy a few beers and only have to do 10% of the work. Also maybe you can code to not change direction till it is out of line by more than say 1m, what I mean is if it has driven 90m with minimal deviation then suddenly a read says you are 0.75m off line without a direction change command you ignore that reading and carry on until your reading says 1m off target. Anyway, this would be interesting and I feel your GPS coordinates would be more accurate than in the city with our tall buildings blocking the signal. Also another idea is to mow in strips up and back and after finished do crossways, that way every area gets hit twice, the chances of a bad GPS read causing a missed section is halved. Of course this will use twice the fuel. But I would weigh up fuel cost to relaxing in the shade with a frothie. Regards, Mick Mick's uMite Stuff can be found >>> HERE (Kindly hosted by Dontronics) <<< |
||||
| Gizmo Admin Group Joined: 05/06/2004 Location: AustraliaPosts: 5012 |
Yeah its clear open sky here, even the wifi works for hundreds of meters. I have about 3 acres I want mowed automatically, I'll hop on the ride on mower to do the edges and around trees. The mower projects I've seen use a combination of good GPS, compass and wheel encoders to get really nice rows. Some of them are using drone flight computers, others using a arduino and a pi for vision based obstacle avoidance. Its the challenge of fine tuning and software smarts to get the best results thats the fun bit. The mower itself is the size of a small ride on mower. Most of the hardware is sorted and I can drive it around with a remote control, but thats boring :) Its a bit over powered, about my jogging speed at full speed, but I will restrict that to a walking pace. I have 3 paddocks to mow, the rest I'll do on the ride on. The mower should be able to revisit every paddock every 2 to 3 days days, so its ok if it missed a bit the last time. It will also need a return to base function if it detects poor GPS, rain or low battery. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 3496 |
Sparkfun have a writeup how they did this with more or less basic components (ESP32, Raspberry Pi etc... G**gle for Sparkfun RTK Edited 2022-10-31 17:43 by Volhout PicomiteVGA PETSCII ROBOTS |
||||
| grunto Newbie  Joined: 07/01/2015 Location: AustraliaPosts: 15 |
iforce2d did some videos on a "Farm Rover" that used path-following that may be of some interest =http://https://www.youtube.com/watch?v=6Ic1B9Mg6i4&list=PLzidsatoEzej82fN6okQjhz7DYPisJYKE]/URL] |
||||
Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 5705 |
Hmmm... wondering about a "line follower" that looks for modulated laser light. It's something a bit different. :) Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| ryanm Senior Member Joined: 25/09/2015 Location: AustraliaPosts: 191 |
I used to do surveying. Differnetial GPS does the corrections on the signal before it gets turned into a position. If you try to do corrections based off the processed lat/long you will pretty much just get random noise that might even be worse than the original signal to the best of my knowledge. You can get RTK kits relatively cheap now (ie $600 for a set). I would recommend going with an ardupilot brain and one of the RTK kits from HERE or EMLID if you want to go cost effective off the shelf. Quite mature system that has been field tested for you. On the diy route there's two options that I see as feasible. Look into software called RTKLIB and try to find two cheap GPS units on aliexpress that are supported. Fairly steep learning curve though. Second you could triangulate position based off the bearing to 2-3 known points around the yard. Either an IR beacon or some sort of visual landmark like an april tag then run machine vision on a raspberry pi with a few cheap USB cams hooked up. |
||||
| ryanm Senior Member Joined: 25/09/2015 Location: AustraliaPosts: 191 |
More crackpot ideas free of charge... If you have a fence on one side you could use GPS for course location then use an ultrasonic distance sensor ranging to the fence to maintain a set distance. This would get you perfect straight rows. Or let the grass grow a noticeable amount (2-3 inches) between mowings. Then you could use some sort of proximity sensor, ultrasonic or infrared most likely, to detect where the edge of the cut grass is. All you would have to do is remote control the first run around the perimeter, then the mower could follow the cut edge of the grass in an inward spiral with a bit of line following code. |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 3496 |
Typical european mowers use fencing. The cheap ones use random mowing, only expensive ones run intelligent patterns. Is there a fence (or barb wire) you can use to align to ? PicomiteVGA PETSCII ROBOTS |
||||
| Page 1 of 2 |
|||||
 Print this page |
