Skid Steering out of control

I did a test run in the nice, sunny weather today with the 2015-March-13 APM Rover software.
The navigation refused to work!!! The rover started, seemed to steer but never aimed for the WP:s. I had to fetch it over and over.
Just to test I swished the connectors for the ESCs and let it run. I thought it would circle but it started weaving and moving forwards. Good I thought, then there´s only parameter adjustments but no. Check:

The Rover started driving at the point where it is in the picture. It was weaving in a very long bend to the upper left corner where I caught it and disarmed it. The track back is when I was carrying the rover back to my car. (Home)

As shown i MP it is aiming for WP1 in Auto mode but the path it is following goes somewhere else.
I hope it is my fault so it can be easily corrected.

My setup is 4WD skid steering with 4 brushed 12V motors. Two Graupner Speed Profi 40R powering two motors each. One for each side.
Piksi RTK kit as the first GPS and uBlox as the second.
The Pixhawk has the arrow pointing forwards. The uBlox to.

It seems very, very strange, to me, that the rover run a relatively straight line forward when I switch left and right controllers. Is it possible that Gyro and GPS give conflicting messages?

What version of the ArduRover2 firmware are you using?
I have validated that the differential (skid) steering works for v2.47. I have not tested v2.48 or 2.49.
What do you have the following parameters set to:
Have you run the rover in the Manual Mode to ensure that it is steering correctly?
Did you run a bench test to verify that when in the Auto Mode the rover will steer correctly to a preselected waypoint that is either to the right or left of the boat?
You might try checking the “reverse” box next to the aileron (roll) bar in the MP radio calibration screen. In some cases it takes reversing the motor controller outputs when in the Auto mode as you have done by reversing the inputs to the motor controllers.


You are always quick to respond. Thanks.
2015-03-13 is the latest version of 2.49
My rover is a ground based 4WD vehicle but I guess it doesn’t matter.
When I use the Graupner receiver the servos are going the right way and the vehicle never tries to go backwards.
I noticed when I came back home the the compass was slowly turning although the vehicle was standing perfectly still. To be sure I recalibrated the compass with success and new values has been stored. The compass is still turning (yaw) in Mission Planner!!!
How can this be possible? Is it a hardware failure?
I currently drive the vehicle without RC. I send the mission and start it with Mission Planner.
SKID_STEER_IN is 0 (disabled)

The servos almost never reverse. Thats also very strange. They may do so indoors when I test but never in the field.
I do not reverse any channel in the Pixhawk. I had an issue some time ago with my octocopter so I learnet how to do it in the transmitter. Thats is for flying. The rover has another setting without any reverse but as written before I currently doesnt use the RC. One of my receivers is at repair and the one I have left cannot sum the signals into a single wire.

Sorry, I had boat on the brain and not rover.
The only time that you want to reverse the steering at the R/C transmitter is when in the Manual mode.
If you need to reverse the steering in the Auto mode then you need to check the “reverse” box next to the aileron (roll) in the MP radio calibration screen.
Since swapping the Pixhawk outputs to the motor controllers allowed the rover to steer straight, then you might want to check the “reverse” box next to the aileron (roll) bar.
Have you verified that your compass is not being affected by the rover’s motors magnetic fields?
You really need to be sure that the rover steers correctly in the Manual Mode using an R/C transmitter and run an Auto Mode bench test as I described in my post above.

OK I’ll hope to do that tomorrow. I can borrow the receiver from my octocopter.
I have the uBlox on the GPS Mast and the Piksi GPS antenna on a higher mast.
The Pixhawk is not very far from the motors but as they are disarmed they provide a steady field, not a rotating field.
I have servo extension cables. I will make a test run tomorrow with the Pixhawk further away from the motors.

Now I have done a lot of testing. Rebuilt the rover to get the Pixhawk mounted further away from the motors. Compass re-calibration removed the problem of constant slow turning although the vehicle was standing still.

It still refused to follow any path today. Tried, once again, to shift the servos but no big difference!!!

Some brainstorming and then I switched the connectors from uBlox and Piksi and the rover navigates!!!

If I correctly understand the functioning of GPS the direction of the external GPS antenna has almost no effect on GPS positioning. Movement is calculated from the satellites only. Right?

If my understanding is correct there is a bug in the Piksi integration!


And please:
For every latest version, write a summery of what’s changed!
I’ll gladly continue testing but I’d love to know what improvements to test.


Basically v 2.49 was a bug fix build:

I’d suggest upgrading the firmware on your Piksi’s. I have a set on my desk that I haven’t had a chance to get to but I know they have been working hard on firmware updates so you should do that first. I would then have the Piksi as the second GPS initially and do a few missions then graph or replay the mission and compare the differences in the GPS positions between the first (ublox) and second (piski). That will tell you if the Piksi is behaving.

Thanks, Grant.

I have a boat, skidsteer with 2 blue robotics t200 thrusters. Everything is flawless in manual mode. But if I put it in auto, or rtl, it heads away from me, or away from the way points alternating left and right motors. As soon as I flip it back to manual, it’s all good. Any ideas?

Try checking the “reverse” box on the Roll bar channel on MP.

I had one of those stupid moments haha turns out I had waypoints saved on it from mission planner and was setting waypoints from my phone in the water and it was trying to go to the ones from mp. Thanks for the response!