Community Forum

Robo Mower

(Elbarfo) #41

I already have built the base package and started working with a couple of nodes. I think I can use several of the existing libraries for my needs. ROS really is brilliant.

Yeah, I’ve been waiting a while for the Sweep now. They sure are very nice for the price. It is possible to do a similar thing with the Lidar Lite but I’d have to brew the control system for it. I understand the Sweep may not be so rugged either…I want to hear more from users first.

So I take it then you’re not using the GPS unit on the Navio? That’s more what I was wondering. I like the Navio’s easy integration with a radio. Would make a lot of things easier. I think the Reach + Navio is really what I need but it breaks my budget :frowning:

The reach units really would simplify things.

(Mjanicko) #42

Hi mr337,

its quite interesting project. Do you think with Pixhawk and Reach RTK kit you would get the same
excelent result?


(Mr337) #43

Hi Mjanicko,

The Reach should get similar results, RTK Fix is RTK Fix regardless of where it comes from.

The Pixhawk should be able to get something similar, but you milage may vary with ArduPilot. Hence why I ended up writing my own and passing on ArduPilot.

Best of luck!

(Esa Attia) #44

HI mr337,

Can you go explain why the ardupilot didnt work like you wanted?
Would really appreciate your insight.

I have a similar project (differential driven platform) and was trying to leverage the stuff already done by ardupilot and just write a few ROS nodes to interface with mavros/ardupilot.

What roadblocks did ardupilot/mavros present?


(Mr337) #45

Hi Esa,

I think there needs to be some clarification. I have gone commercial with my mower project. This changes the perspective of how these things look.

ArduPilot is great for a lot of things. It just didn’t align too well with the requirements I had such as support cadence, implementation features, etc. We have pretty specific stuff that we need to bed addressed in regard to land vehicle (rover), that writing our own implementation seemed like the only path forward.

Here is an example of digesting 2D LIDAR data.

Digesting the data wasn’t a problem. The output of the data touched every part of our subsystems since it was related to safety. IMO this would be a huge undertaking with using a system like ArduPilot or the effort of maintaining a fork and backporting patches which would require more work as the fork diverges.

I can’t comment on the ROS + ArduPilot bridge as it was easier to roll our own.

(Esa Attia) #46

That’s fantastic ! Wish you all the best getting mowbotix going.

I can now understand why you are not using ardupilot. Thank you for the explanation.

Since ardupilot is not critical to your system. Are you still using RP3+ Navio2 + Reach in your setup ? Technically you can use any Single board computer + Reach to do the same thing since you are only using ROS.

How is the RP3 holding up in terms of processing capabilities?

Is the accuracy of the Reach module still good enough for what you need to do? Do you get a constant fix when attempting to do RTK while moving (in clear open sky of course) ? or does it fluctuate to float constantly? Di you find weather conditions to be affecting the RTK result ?

Are you using the IMU brick for compass ?

(Mr337) #47

In the video we are using a RP3 and have moved to our own IO interface for low level/realtime stuff.The Navio2 worked great for development and unless you want to do it yourself, stick with the Navio2. It saves a lot of work! RP3 is great for prototyping, but we have moved from it since we hit the computing capacity of it already.

We aren’t using the Reach, more info here Robo Mower.

We are tinkering with the IMU brick from tinkerforge. Still are exploring different options in that space.