Sometimes when I start the Slamdunk, the SLAM axes are in a strange configuration regarding the world frame - a constant offset of some degree in the X, Y, and/or Z axes. (So I’ll move the Slamdunk forward, and it’ll think it’s going 30 degrees to the right, for example.)
Even when I try to use a rosservice call to restart SLAM, hoping the axes align with the world, the Slamdunk reference frame stays skewed. If I knew what the offset was, I could give it an offset (x,y,z) to compensate, but the offset changes every time (though it only happens sometimes - othertimes there’s no problem).
I read in another thread that I should start the Slamdunk in an area with good features. So I am only using the Slamdunk in an area where it sees good features upon startup.
Similar - there is a constant offset for the ultrasound distance, which also changes every time. Ultrasound jumping & offset problems
Edit - I ran “sudo diagnostic”:
01-linux-version ok [system] Linux version
02-memory ok [system] Memory close to 2GB
03-linux-thermal ok [system] Linux thermal zones availability
04-cpu-governor ko [system] CPU governor
05-linux-sanity ko [system] Linux sanity
20-magnetometer ko [sensors/i2c] Dev ID of AK8963C magn on /dev/i2c-0 (Result: 0x0f)
21-barometer ko [sensors/i2c] Ping MS5607 barometer on /dev/i2c-0
22-codec ok [sensors/i2c] Ping ADAU1772 codec on /dev/i2c-0
25-ultrasound ko [ultrasound] Ultrasound Result: KO Distances: ""
30-camera ko [camera] Cameras acquisition
31-camera-freq ko [camera] Frequency Result: ms (33.33 ms) diff=-inf%
32-camera-conn ko [camera] Camera connection: Left is unknown and Right unknown
40-imu ok [imu] IMU Result: 15935 lines (16000) diff=0.41 %