Copter-4.1.0-beta4 release for beta testing

iampete · June 19, 2021, 9:56am

Yeah, looks like a script error. If motors matrix has not init it will never setup the motor outputs as Dshot. If you can post the script I will see if I can spot anything. The other possible issue is the assigning of motor outs, the scripting back-end will not auto assign motors to output functions 1-6 as the other back-ends will. You have to manually set the outputs 1-6 to motors 1-6.

Lorenzo_Pessah · June 19, 2021, 10:03am

Here is my script (last post)

iampete · June 19, 2021, 12:11pm

Ok, just read your post more carefully. The script was only failing because you did not have frame class set to scripting in that test.

I suspect is just a matter of boot order. The script comes up after the auto blheli check is done. This is the check:

github.com

ArduPilot/ardupilot/blob/39c96f63c8c844193dedf13c93014913d8075975/libraries/AP_Motors/AP_MotorsMatrix.cpp#L187


    // convert output to PWM and send to each motor
    for (i = 0; i < AP_MOTORS_MAX_NUM_MOTORS; i++) {
        if (motor_enabled[i]) {
            rc_write(i, output_to_pwm(_actuator[i]));
        }
    }
}


// get_motor_mask - returns a bitmask of which outputs are being used for motors (1 means being used)
//  this can be used to ensure other pwm outputs (i.e. for servos) do not conflict
uint16_t AP_MotorsMatrix::get_motor_mask()
{
    uint16_t motor_mask = 0;
    for (uint8_t i = 0; i < AP_MOTORS_MAX_NUM_MOTORS; i++) {
        if (motor_enabled[i]) {
            motor_mask |= 1U << i;
        }
    }
    uint16_t mask = motor_mask_to_srv_channel_mask(motor_mask);


    // add parent's mask

That will return 0 until the script has run. Not a huge amount we can do about that, we could re-init blheli after the script init. But I think on the whole it is not too painful to have to manually set the motor mask.

Lorenzo_Pessah · June 19, 2021, 12:30pm

can I improve the script in any way?

iampete · June 19, 2021, 12:36pm

Not really. If your feeling keen you could have it work out and set the manual blheli mask, so that it would work correctly after a reboot. but that would be quite a bit of lua just save the manual setting of one param.

Lorenzo_Pessah · June 19, 2021, 1:00pm

I have also noticed that if I reupload the parameter file, outputs remain in pwm, I have to rewrite the manual mask and reboot in order to make it working

Quadzilla · June 19, 2021, 6:41pm

ArduCopter V4.1.0-beta4 (37070bfe) the first 30 secs I did have some shakes in stabilize then they went away, switching into GPS the same shake returned, "not seen this " overall stock tuning seems ok for this size mini beast. I play with the pids next. Tested ground effect control gave some hits to the Tarmac but was able to get some of the leaves off the driveway.

MartyMcFly · June 19, 2021, 7:49pm

that is an awesome design with great potential,what size motors are you running on it please

Quadzilla · June 19, 2021, 8:27pm

Hi martin, it’s designed to use 2000 kv or 1800 kv will work with a x or quad just add the bottom unit. this is a 1/4 scale version above.

MartyMcFly · June 19, 2021, 9:08pm

I could have fun with that,put a nice rotating lidar on it set at say 1meter and well set up could fly all day flying at wall’s hedges etc

andyp1per · June 19, 2021, 10:06pm

I really, really recommend you don’t try with stock tuning. Every copter is different and stock tuning is more likely to disappoint than not. Auto tune is your friend.

Quadzilla · June 19, 2021, 11:03pm

Sorry Andy I been using stock tuning for 5 years with a lot of builds. 90% of the time they fly prefect so no worries my friend. if i do have an issue i manually tune. My test air space is full of trees. However by all means Autotest is very cool and thank you guys for what you do mind blowing.

Derek · June 19, 2021, 11:20pm

Is only 1 down facing rangefinder allowed? Say one lidar is good from 40m to 1m, and a second is good down to 0.1m (but maxes at 6m outdoors). Is it possible to set them up so it uses one then the other?

Quadzilla · June 20, 2021, 7:40pm

I found my jitter. I ended up turning down the PID " possibly more as i hear something after looking at the video " and added 45 degrees tubes. GPS stick control seems a little slow and BHL32 bit motor test is not working. however i was able edit the esc settings. Beta-4 is feeling good.

rmackay9 · June 20, 2021, 11:20pm

@Derek,

Yes, it is possible to use two downward facing lidar but it will only use one at a time. The is the logic on how it chooses which of the lidar to use:

use the first lidar by default (e.g. RNGFND1 will be used before RNGNFD2)
if the first lidar goes out of range the 2nd one will be used. Note the 1st lidar driver reports whether it is out of range by checking the distance vs RNGFNDx_MIN_CM and _MAX_CM

I must say that I’ve never tested this feature personally but Tridge has so I hope it works! Please tell us how it goes.

Derek · June 21, 2021, 5:48pm

That got me thinking, I just received a few Thoneflow 3901uy optical flow sensors, and they are quite small. I’m a little worried about a piece of grass or dirt sticking to the lens and losing the sensor data. Is it possible to use 2 optical flow sensors, 1 per lane ekf3? I see the current setup is 1 lane per imu, but what about 2 copies of each imu and alternate the optical flow sensor? So if you have 2 imus, there would be 4 ekf lanes (or x number of lanes per IMU) Sounds like a lot of work to code tho. The sensors are inexpensive and it would be nice to add a backup for reliability

rmackay9 · June 23, 2021, 12:08am

Hi @Derek,

We don’t support multiple optical flow sensors yet I’m afraid. It has come up a few times although normally the request was to have them pointing in different directions. What you’re describing is “sensor affinity” for optical flow which is also a good idea.

Derek · June 26, 2021, 2:58pm

Trying to think out how I want my multirotor set up. I seem to be running low on serial ports. I’ve got a pixhawk 1 (both a 1m and a 2.4.8, I could use either). Got the 4.1.0-beta4 firmware. Currently the GPS, telemetry radio, and the OSD use serial UART, the mag is on i2c and so is a small LCD screen. Multirotor is mostly done building just want to add some additional sensors (it has successfully flown). I’d like to use 2 lidars (TF mini plus) which are set up as UART but I think can be switched over to i2c, but that raises a question how much can I add to the i2c bus before problems arise? I’d also like to use an optical flow, Thoneflow3901uy which can only run on serial UART iirc. Am I overloading doing the following?

4 i2c- Mag, LCD screen, (2) TF Mini plus lidars (1 down, 1 front).
4 serial- GPS, telemetry radio, OSD, optical flow

I’ll probably want to set up the EKF3 lane switching right away, right? Going to be flying over corn and landing in grass. I’ve read the corn messes with the lidar.

rmackay9 · June 28, 2021, 3:12am

@Derek,

I don’t have much advice except that the oscillation issue flying over a corn field (aka “Dolphin flying”) has been somewhat resolved - at least the performance in -beta5 should be as good as it was in Copter-4.0.

sergbokh · July 23, 2021, 8:02pm

Hi @rmackay9, please advice how could I find the reason of the unexpected climbing / estimated altitude loss. I’m using GSF (no compass)

My steps were:

Wait for good GPS fix, home point set.
arm in AltHold, takeoff smoothly to 2-3 meters alt
then the copter did fast climb for about +2-3 meters.

In the log I see that ArduCopter thinks the altitude started to decrease (that wasn’t the case really):

so it does increase desired climb rate.

But I didn’t found any reason of why it thinks so. There was no Alt decrease nor in Baro nor in GPS logs. No decreasing as per Accel too.

2021-07-23 20-37-29.bin (865.5 KB)