Georacer
February 11, 2026, 7:59am
1
Attendees (unique) : 7
UTC0700
master ← andyp1per:pr-rng-dronecan-low
opened 08:25PM - 24 Sep 25 UTC
The current code means that you will always get "No Data" (which is fatal) rathe… r than too low (which happens when you are sitting on the ground).
Peter : We do have an existing pattern to write an invalid value when the status of the lidar measurement is invalid.Andy : Tridge wanted to send the last valid value.Andrew : For out-of-range-close, we don’t do that.
I would prefer we were sending the last valid value, or the current value and have the front-end manipulate it.
A : Needs testing. You can use the frame quad-can to test it, it has a simulated CAN rangefinder.
UTC0715
master ← andyp1per:pr-fence-AGL-update
opened 05:59PM - 29 Nov 25 UTC
this allows for fence altitudes to be above ground level, absolute, origin or ab… ove home (the default and current behaviour)
This is a rebase of https://github.com/ArduPilot/ardupilot/pull/29761 which I could not push to
P : I am concerned that if we land on a mountain, reboot and set the origin there, we won’t be able to take off again because it will be violating the lower fence threshod. Because it will be too high above Home.
I’ll write a test for it.
UTC0723
opened 06:01AM - 16 Jan 26 UTC
## Bug report
**Issue details**
We are testing Bosch BMP581 as a replacement b… arometer on Pixhawk 6X due to ICP20100 and BMP388 being discontinued.
It seems like
1. BMP581 works fine on I2C2.
2. On I2C4, we see intermittent bad samples that show up as large spikes in pressure or baro altitude. Logic analyzer suggests occasional I2C glitches. This seems like bus reliability or recovery behavior rather than a sensor algorithm issue.
<img width="813" height="265" alt="Image" src="https://github.com/user-attachments/assets/51068d07-2f27-4fd0-9251-fb634ef2f821" />
baro0 (i2c4, BMP581 glitches)
<img width="770" height="328" alt="Image" src="https://github.com/user-attachments/assets/00904aef-9db4-45f5-bb2a-5e95dd4f2155" />
Baro1 (I2C2,BMP581, No issue)
<img width="1283" height="555" alt="Image" src="https://github.com/user-attachments/assets/e3718308-2979-4f7d-ac88-2d59b1f3a647" />
**Version**
Master
**Platform**
[ X ] All
[ ] AntennaTracker
[ ] Copter
[ ] Plane
[ ] Rover
[ ] Submarine
**Airframe type**
Copter
**Hardware type**
Holybro PIxhawk 6X
**Logs**
[Log ](https://drive.google.com/file/d/1VpvUOkHgXlUoMBhySO-Vg0Qk_fINYPsm/view?usp=sharing)
A : We have a filter library, which could be looking for too great differences.
We could try the register twice, provided they don’t get cleared after each read.
G : The problem could be electrical.A : Most likely, but it wil be common, so we can do our best on our end as well.
UTC0732
master ← IamPete1:Q_LAND_FINAL_ALT_doc_fix
opened 01:05PM - 09 Feb 26 UTC
Theses were missed in https://github.com/ArduPilot/ardupilot/pull/26025. I also … found a reference in the payload place script. I have not tested the script.
Merged!
UTC0734
master ← peterbarker:pr/norm-stuff
opened 03:44AM - 08 Feb 26 UTC
Where `sqrtF` is double, this prevents promotion of floating point values to dou… bles just because we are calling `norm` on them.
```
Board AP_Periph antennatracker blimp bootloader copter heli iofirmware plane rover sub
CubeOrange-periph-heavy -40 *
Durandal -64 -72 * -88 -112 -120 -80 -96
Hitec-Airspeed * *
KakuteH7-bdshot -48 -48 * -72 -96 -96 -56 -72
MatekF405 0 0 * 0 0 -8 0 0
Pixhawk1-1M-bdshot 0 0 0 0 -8 0 0
SITL_x86_64_linux_gnu 0 0 0 0 -4096 0 0
f103-QiotekPeriph 0 *
f303-MatekGPS 0 *
f303-Universal 0 *
iomcu *
mindpx-v2 0 0 * 0 0 -8 0 0
revo-mini 0 0 * 0 0 -8 0 0
skyviper-v2450 0
speedybeef4 0 0 * 0 0 -8 0 0
```
<img width="3734" height="865" alt="image" src="https://github.com/user-attachments/assets/922e6c54-9658-4f0a-8115-85d227b889cc" />
Typical change, 64-bit operations on the left, 32 on the right, and note no longer using `vcvt.f64.f32`
<img width="4796" height="3616" alt="image" src="https://github.com/user-attachments/assets/22cfde12-68d7-4607-a24d-aa7e7cebb88d" />
Merged!
UTC0737
master ← peterbarker:pr/correct-binlog-rotate-issue
opened 10:39AM - 04 Feb 26 UTC
when a test fails we can move the log files to a permanent storage location for … later perusal.
However, we do this moving of log files after rebooting the autopilot. If we are logging disarmed then we will be moving the currently open log file, which is not good.
This can kill random tests that try to get the most recently open log file by globbing the logs directory and then indexing by -1 - because the list can be empty!
Merged!
UTC0739
master ← peterbarker:pr/stop-moving-logs-on-test-failure
opened 10:07AM - 04 Feb 26 UTC
tracking down the logs on test failure is just painful.
I've been setting the… environment variable for a long time and it works well
Merged!
UTC0740
master ← peterbarker:pr/remove-optflow-srtm
opened 03:26AM - 29 Jan 26 UTC
A : Let’s hold off for Randy to have a say.
UTC0743
master ← ZeroOne-Aero:pr-ADIS16507-260126
opened 08:40AM - 26 Jan 26 UTC
Fixed issues related to ADIS16507 IMU data processing that caused incorrect gyro… scope and acceleration values:
<img width="554" height="463" alt="image" src="https://github.com/user-attachments/assets/d49ace09-8688-4f4b-b293-1536d5f87962" />
- According to the if in line 235 , both 1200 and 1000 can obtain the value of REG_DEC_RATE_1000Hz(REG_DEC_RATE 1).
- However, the values used in the calculation for lines 562 and 563 were incorrectly assigned as 1200, while the calculation formula in the Decimation Filter manual is 2000/(DEC_RATE + 1)=1000.
<img width="927" height="1266" alt="9cf0e8f563083d70473da8741edfa4d8" src="https://github.com/user-attachments/assets/45e4502c-ab90-4848-bf33-796fd69c6ef6" />
<img width="772" height="1039" alt="1e090a90e6ccd4aaf8f0e3fca3d04813" src="https://github.com/user-attachments/assets/0c9e893c-4714-4252-8fcf-646f669d332b" />
- Incorrect assignments led to abnormal sensor calculations and abnormal refresh rates.
Merged!
UTC0745
master ← andyp1per:pr-linux-serial-led
opened 11:29AM - 19 Jan 26 UTC
## Summary
- Add modular SPI-based serial LED driver for NeoPixel/WS2812 LEDs o… n Linux boards
- SerialLED_SPI is a standalone class that can be composed into any RCOutput implementation
- RCOutput_RPI provides the RCOutput wrapper for Raspberry Pi boards
- Uses 8-bit encoding at 6.4MHz SPI clock where each WS2812 bit becomes one SPI byte
## Features
- Supports MODE_NEOPIXEL (GRB) and MODE_NEOPIXELRGB (RGB) color ordering
- Up to 128 LEDs per channel
- Uses /dev/spidev0.0 (GPIO10/pin 19) on Raspberry Pi
- Thread-safe with mutex protection
- Automatically enables channel 0 when HAL_LINUX_SERIALLED_ENABLED is defined
## Hardware Setup (Raspberry Pi)
- Connect NeoPixel DIN to GPIO10 (SPI0 MOSI, physical pin 19)
- Enable SPI in /boot/firmware/config.txt: `dtparam=spi=on`
- Set NTF_LED_TYPE=1 (NeoPixel GRB) or 128 (NeoPixel RGB)
- Set NTF_LED_LEN to the number of LEDs
## Test plan
- [x] Build for pizero2w board
- [x] Hardware tested on Raspberry Pi Zero 2W with WS2812 LED strip
Merged!
UTC0746
master ← peterbarker:pr/covariance
opened 10:03PM - 10 Feb 26 UTC
This notes down that an invalid covariance matrix is marked by a NaN value in [0… ][0].
I've verified that we assume this in ArduPilot already (or don't use the field).
Merged!
1 Like
I would rather have the measured value as it may be of use to a script or for log analysis. Eg rangefinder that works reliably up to 15m but can provide data up to 25m if the conditions are right.