When a robot can't tell you why it dies

GoPiGo
,
1

GoPiGo3-Dave Was Lurching Off His Dock and Dying Quickly and Quietly

Took me three boots on the dock to understand what was happening:

  1. I shut Dave down this morning to make a backup - (removed from dock, turned the battery switch to off)

  2. Returned Dave to Dock, pressed the GoPiGo3 “On” button

  • Green light flashed until processor booted up
  • Processor performed all startup functions including starting Docker and “ROS 2 dave_node”
  • GoPiGo3-Dave lurched forward to end of docking contacts and immediately died, no yelling, no logging.

  1. I repeated step 2 twice more with exactly the same result.

  2. Realized I did not turn the battery switch to on, so dock was powering GoPi5Go-Dave until the moment he undocked.

:person_facepalming:

2 Likes
2

picnic

You need a second set of switch contacts so that Dave can monitor the battery switch state.  With the associated log file, of course.  (and a ROS2_Dave “is the switch on?” node too.)[1]
:wink:

Seems like this is getting too complicated for its own good.  Unless you’re planning to send Dave to Mars. (!!)

=============== Footnotes ===============

  1. If it works, maybe you should back-port this to Carl?
2 Likes
3

The whole power system is such a kludge, not going to “rearrange the chairs …”. I do not need to turn off the battery, only shutdown -h and press the GoPiGo3 power switch to finish the power off.

Carl’s “juicer” program has a heuristics rule set that is smarter, and Carl only has one power switch - no battery switch on his 8x NiMH system.

2 Likes