|
|
|
|
|
|
by thelean12
1883 days ago
|
|
|
As someone who has written robot software, it's not as bad as you might think. You do a LOT of testing before hand, and you do system checks before committing to something. I was always more worried about sensor failures that my software doesn't pick up on. Then again none of my robots ever cost $80 million dollars and has landed on Mars... |
|
|
I've been doing industrial automation with commercial robot arms, where testing is mostly functional and there's very few system checks. You're basically jogging the robot with the teach pendant, recording points, and reading digital IO. There are minimal system checks, simulation is mostly to make sure the cell CAD layout is reachable and less to verify that things are working correctly. Version control is completely offline, there's not even an "undo" feature to revert a touched up point unless you underwent a tedious and slow backup procedure.
That said, like you, none of my robots cost more than 6 figures, and they're all permanently anchored to terrestrial steel and concrete...