A brief serial hiccup on an environmental chamber used to be enough to stop a run. Because status polling was treated as fatal, a single failed read could halt the workcell mid-experiment.
Status polling is now non-fatal: if a read fails, the driver falls back to the last known telemetry and keeps going, while command-level serial faults trigger an automatic reconnect attempt. Persistent polling problems raise a Slack alert so the team knows a chamber needs attention without a run having to fail first.
