Ardupilot Lawn Mower

I want to make my rider lawn mower autonomous, however, I am more confused than ever about the hardware I need to make this work. I would like to be able to have pretty accurate GPS along with the ability for obstacle avoidance sensors. I think I need the rover firmware, but choosing hardware seems to be more difficult. I see a lot of people use the Pixhawk series, but I noticed something about it being outdated and or discontinued as well? What would be the best somewhat newest model that would allow for growth and is not immediately useless after a short amount of time? I read the guides on ardupilot, but I don’t know if I’m just flat out missing something or what, but everything does not seem as simple when choosing hardware, and there are so many options that it doesn’t make it as easy to choose. Also, I would like to be able to view the onboard camera as wellI, and the ability to take control with my rc controller (Radiolink AT10ll) if need be. Also, I would like to use Rover 4.0 or at least the newest firmware version if possible and I plan to use Mission Planner by the way.

Would the bundle below suffice or would I need something else? Telemetry radios, etc? Also, if it will work is there any place in the US I could order that to have it here much faster? Any advice is greatly appreciated. Thanks.

shop.holybro.com/pixhawk-4_p1089.html (Pixhawk 4(plastic case)&GPS (UBLOX NEO-M8N)&PM07)

I have no experience with ardurover, but for sure you need RTK to get the required accuracy to mow your lawn. Normal GPS, no matter how good the quality, will only be accurate within a few meters. I don’t know if you have young children, but for sure I wouldn’t have an autonomous lawn mower doing its job with children around. No matter how good the object avoidance.

I have just read your comments and unfortunately the reality is that you need to spend at least three months to do some research to understand how the mission planner application works.
The reality with any of this technology is time and tears.
When I first started I asked all the key stakeholders for help and didn’t really get a good response. The only way that I could make my application successful was to keep on reading and keep on reading and hoping that the equipment that I was purchasing would actually work.
What works for me is the pixhawk black cube.
Mission planner took a few nights to understand the configurations but again it comes back to your background of electronics and a solid understanding of software configurations.
Because this is not my full-time job it takes a lot of concentration and dedication to get to the point where you can build a platform that you can plot waypoints and finally get your platform to move in the direction that you anticipate.
I found the sabretooth dualdrive to be reliable.
I found Ardusimple to be reliable down to centimetre accuracy but to show you how it all works happy to do it to teach you how it works it would take too long.
What I really find frustrating within my project is it one can go out and spend hundreds of dollars only to realise that it’s a waste of money it’s all well and good to buy these electronic devices and components but if they don’t work the providers have a case to answer proclaiming what does or does not work ends up being the cost of the end user.
So it comes down to the basic fundamentals, be prepared to spend 3 to 6 months in learning then be prepared to spend 3 to $10,000 to build your project but we’re talking about a project that could be productive in real life and in real time the fact that we can go out and buy a module to have several layers of satellite systems for a few hundred dollars is incredible this part is the most exciting part of the satellite GNSS system.

1 Like

The Pixhawk 2.x flight controllers are outdated because they use the STM32F4 MCU, which has limited performance. The key is to look for a flight controller with 2Mb flash memory, as that will get you the full feature set of ArduPilot. See Choosing an Autopilot — Rover documentation for a list of supported flight controllers.

Any flight controller with a STM32F4 or H7 processor and 2Mb of flash memory will last a long time.

That would work fine. There should be some US resellers around, but I don’t know of any off the top of my head.

Also, see Big GPS Round Up for a recent review of GPS modules

Hi,
I’m a first master student doing a project around autonomous driving in a tractor. We have a tractor with all the hardware installed. But we can’t figure out the right PID-settings for turn rate. We are using an old tractor (that already has a PID installed apart from the one used for ardupilot (pixhawk)). So we need a good understanding of what the PID of ardupilot does. The ardupilot site gives not much explanation about this. What if the PID is zero? What does it sends then to the tractor? Is the PID of turn rate coupled to that of speed 2 throttle because turn rate is dependend on the speed, no? Has anyone some PID settings used for a large vehicle that we can use as a starting point? Because using the ardupilot instructions on tuning the pid for turn rate won’t work for us. We are really desperate… Thank you!

We have large agricultural rovers and have gone through this also.

If your existing steering controller controls steering rate and performs adequately then you may just need to use the FF (Feed Forward) term in the Ardurover steering controller. If all of your control loop constants are zero then the output will be zero. The FF term contributes a fixed proportion of the requested steering rate to the output.

The L1 navigation controller outputs a lateral acceleration request in its attempt to stay on track. This is converted to a steering rate in rad/s. The code that converts the steering rate to an actual servo output using the PID + FF constants is found here. You will see that the ff term is simply added to the output. The ff term is calculated here and is just the product of the ff constant you provide and the current target. Thus a ff constant of 1 will give a full range output on the steering when the target steering rate is 1 rad/s. The other PID constants are applied similarly but are calculated based on the error (the difference between the desired and actual steering rate). Hopefully this helps.