Easily within a foot, so long as it’s well tuned and provides repeatable path following.
I routinely do so with my big mower, though I tend to keep a wider berth on big/expensive obstacles because I make such frequent/experimental updates in the interest of testing. But the latter is more of a “me” problem than a system issue.
A word of caution on large/expensive things from Alibaba/AliExpress/BangGood/etc: keep an eye on the shipping costs, need for a freight handler, and/or need for a customs broker/expediter. Added fees can quickly overcome the “good deal” you think you’re getting. I know @swebre had a fairly negative experience in that regard a year or so ago.
I was thinkig about the same, but fully electric. My biggest concern is route planning around obstacles and cut quality around them. Especially in confined spaces with many obstacles it might be an issue, MP route planning is far from optimal for cutting application. Unless something has changed in the last year.
I find that in really confined spaces even something simple like a small WP Circle becomes a problem. The smallest circle of waypoints you can do is 2m radius. If you want smaller than that, you have to create it manually. A small mower like that could mow a much tighter circle around a tree. I have been thinking about creating an Excel spreadsheet that would take a manual pattern you created and recenter it around a new GPS location, then you would just paste in the new locations to a waypoint text file. A manual pattern would only have to be created once. If there is already a way to do this that is built-in to Mission Planner, let me know. The pattern might be a series of passes next to an object at angles, indexing around the object. You could even move letters of the alphabet around, if you wanted to write something in the grass.
Route should be changing the angle, so it does not go over and over the same path (and making turns) destroying the grass
It should be making at least 1 pass around the perimeter of the obstacle
Zig-zag path should be changing the direction at the obstacle, but also do another zig-zag path behind the obstacle. In combination with multiple obstacles path planning might be quite complex
Having a logic for route re-planning in case unforseen obstacle occurs
Making turns on 4x4 needs to very well be finetuned, otherwise it rips the grass (seeing this challange on the factorymade robotmowers)
All in all, I see no other way but to build fully custom path planning program just as Christopher_Milner did.
@Muse I don’t disagree with anything you mentioned, but creating the entire Mission Planning software is a major undertaking that takes a skillset I don’t have. I am more focused on making what we have work so I can mow the grass right now. I think it grows a couple of inches every day. For my purpose, I am still thinking about the little Hybrid mower that has an electric drive motor on each wheel and the gas engine to run the blade. The engine also charges batteries that power the drive motors. Those are tough mowers that can mow over-grown grass pretty well on acreage out in the country. They are small, but would still work for the close mowing around obstacles and I don’t have to build it from scratch.
Raspberry Pi Camera and v4l2rtspserver to stream it at 1080p.
Use any RTSP viewer, including VLC and others.
My mower is all electric, and I need to revisit the camera - I wound up using a long pi-camera cable, and they’re unshielded, so the camera works great when it isn’t moving, but RFI from the brushed DC motors causes some hash on the video feed (it also wrecked the compass, thus RTK Moving Base for YAW)
So the same seutp might work fine with a shorter Pi Camera cable, or better shielding or less RFI.
If you need another camera, use a Pi Zero + camera and you can operate it as a USB-Ether-gadget and plug it into the main companion computer and now you have two video feeds.
The first USB-gadget is easy, doing a second one takes a little more work with IP address allocations, but you can then do a few more if you wanted.
Hi Tom, Thanks for your advice and comments. So I came across the following problem that may be of interest to others.
Basically every time I would turn on the stream for the USB webcam which was plugged into the RPi, the RC transmitter would start beeping and on mission planner the gps would become erratic and I would have problems talking to the on board Raspberry pi. It turns our that the particular webcam uses the USB3 protocol and this protocol uses the frequency 2.5ghz which ends up creating interference for devices using 2.4ghz, like the Wifi Chip on the RPi and the transmitter. The higher the camera data rate the more interference created. I did buy this web cam as part of a drone kit so this was a surprise to me.
I looked at the pi camera and my thoughts are that if I needed really good latency I think this would be my next move. For now though, I am going with a 5meg dahua cctv camera which talks through a modem to my network, I can ip into it using a browser or vlc on any device. When you select sub stream 1 and reduce the VLC buffering value to 0 or very small (maybe 10ms), I would guess the responsiveness is around 0.3 seconds. This is fine for my lumbering machine. I tested a 5 year old dome cctv camera and it was worse, so the newer camers do have better processing power or something that makes them better
Design methodology wise, I like the idea of not putting camera traffic through my companion computer and so give it its best chance of doing what I really need it for which is telemetry using dronkit, mavproxy (havent really implemented this yet )etc I am trying to keep my systems separate, so that when things go wrong it is easier to troubleshoot.
@SJohnson
Hi Steve, I think I read in this post or some other that you were looking for an emergency cut off switch for your mower. I found the equipment below (you would use this to control a bigger relay/contactor) There are lots of options on the website but the equipment is industrial in nature. I would assume the transmitter only transmits when the buttons are pushed so you would not end up with interference for your existing telemetry equipment, only when you want to kill the machine…and 4 channels so you could select what you killed
Steve, this is the actual manufacturer of the mower I purchased and I’m assuming it is something like you’re looking for. Most of them you find have been rebranded. It saved me around $250 buying direct. I spent $1710 for the upgraded Kohler engine and a few spare parts, shipped to Ohio. Tracy Su is the contact.
So happy to have found this thread. I am also making a robot mower and struggling with hardware and design/architecture choices. Picked Mission Planner and ardurover for software. Yuri has been so helpful to get the cheaper RTK board LC29HEA tested out. Now I’m doing the remainder of the design. Mine is an old 24V electric robot mower that I’m re-doing the controls. Also, budget is an issue. Good news on mine is that there are few interfaces: 2 drive motors, 3 mowing motors, 2 obstacle sensors. Major hardware choices are now: flight controller, usb radios, brushed motor controllers. Yuri’s testing of the LC29HEA RTK board has locked that in for GPS. From my reading, sounds like the Matek H743 wing is a good (budget) choice for the flight controller?!? Holybro 900mhz at 100mw is sufficient for good range (I’m a ham, so 433 is also an option). Motor controller may end up being the Sabertooth 2x12 at $65. Later I’d like to add sonic obstacle sensor and camera. Anyone have comments on hardware choices?
Holybro also makes the budget friendly Kakute H743-Wing flight controller. If you purchase one, I would get one with the headers already soldered in place unless you are comfortable with soldering on a new loaded flight controller board. The headers let you plug your connections into the board without soldering each one in place. I just bought a Kalute H743-Wing from Holybro with headers for $130 an it will give me the same performance as my more expensive Orange Cube that I have on another machine.
Awesome, Steve. Thanks. Frankly I was between the Matek H743 wing and the Holybro kakute h743. Both are currently at about $75. Holybro is selling the kakute without the pins soldered at the same price as with the pins soldered. And it does look like a lot easier install on the kakute. If anyone could comment further on one vs the other for this mower application, I’d appreciate it. Now, I’m definitely leaning to the kakute. at $72.99. Already soldered pins, 7 uarts, 14 PWM, 1M/2M memory. Can’t see any features that I should have that are not there. For now, or the future…
Unfortunately, I am well behind the curve on knowledge. All mine is ancient. I started as a software engineer for NASA doing flight software, but a l.o.n.g time ago… More recently I did a segway from scratch, with some similar characteristics to the mower I want. My RC knowledge is back when individual channels were in vogue and having a ham license was a huge benefit.
Hey @SailorK, welcome to the crowd of mower builders. We all try to help each other and we all bring an experience base to the table. Your electric mower sounds very interesting! Sounds like it is all electric? A lot of the newer ones are hybrid, with electric drive and a gas engine driving the blade. I am also interested in the LC29HEA RTK board for GPS. I’ll have to search out the information on that approach.
Yes, mine is all electric. $50 at a yard sale. I built a 32Ah 24V 8S LIFOPO4 battery and it gets about 6 hours on a charge. Enough to keep up on an acre with random cutting. Probably easily do 2 acres if I get the ardurover controls sorted out.
Check out this thread. Yuri is super helpful, as are the others on this thred: RTK gps on a budget (Quectel LC29HEA)
@SailorK , if @Swebre2023 holds another Mowstock, I had access to most of the mower/rover braintrust for a couple of days - really accelerated my journey.
If you were with NASA/Houston, let me know - am somewhat close and can show you what I did. @SJohnson came over for a day and helped me tons.
I’m learning from a firehose. No, I was attached to Goddard, Greenbelt MD–did flight software on the Hubble and a lot of R&D for Landsat. Am now in Florida with super aggressive grass in the summer. I am taking baby steps. I have a $20 tiny tracked Amazon robot I’m going to get moving first, then built a 2’x2’ tracked rover with the remains from my segway project (two 500W motors, a 2x32 sabertooth that seems to have failed) and finally the mower. When I did a sim in MP, I found it easy to make a mistake and send the “mower” to the middle of the house. I don’t want a 40# mower with spinning blades going out of control, so I’m going to use the big tracked rover with full controls and use it to map the yard and live test all the mapping, prior to moving the controls en masse to the mower. Trying to do all this on a budget. My big tracked rover is a $30 aluminum tube frame and 3d printed parts: corners, wheels, track.
Ah - was at Goddard during late Gemini missions as a kid.
Keep us posted; upload params and logs when able, and I’m particularly interested in workflow- getting from geo references to obstruction free missions…cd
Quick question for the group on best practices. Not sure if this is the best place for the question. My mower has a lift switch on the front wheel, which is designed to cut off mowing if the robot is lifted. I’d like to keep that feature. What is best practice? Brute force a relay on the cutting motors to cut power or go through an interrupt in the Flight computer? Similar question about obstacle avoidance–there are two bumper switches on front and rear bumpers. Would like to add either lidar or sonar obstacle detector.
Great Question. We need a forum for operational running of these beasts. This forum is very build/fix but questions operational safety, cut out switches, not so much.
I cant speak to best practice - but the less “smarts” between your lift switch and the power/drive/motor/blade the better id say. Key also is how to manage short duration lifts (wheelie) should it restart and continue (scary !!)
Im building up a watchdog process - lua script
Bumper bars arent bumped → HOLD and stop ICE
GPS Fix is 6,6 so solid RTK Fix - so the mower isnt mowing gardens : Hold until 6,6
RPM on the ICE motor on the blade is > 0 - not stalled out : Hold and cry fo help
eStops havent been engaged : Stop ICE and cut all power
camera hasnt recognised a person in front of mower : HOLD
Im thinking the Stop moving logic is easy. The restart is harder, as 300kg mowers leaping back into life is not cool.
