diagnose-system

To map robotics system failures to formal problem representations, execute the minimal configuration, classify failures into categories like perception or planning, and link each failure to formalism elements. This process clarifies why a minimal system fails and helps refine research hypotheses.

To classify robotics experiment failures using wandb logs, gather execution results via wandb APIs and apply a taxonomy classification. This categorizes failures into perception, planning, execution, physics, or spec categories to inform your formal problem representation.

Your minimal robotics system fails during the diagnose stage due to issues in perception, planning, execution, physics, or spec categories. Automating experiment runs and analyzing logs identifies these specific failure points to map them back to the formal model.

Yes, you can automate bash experiment runs to test a minimal robotics configuration. The process locates and executes the minimal system setup using Bash commands, gathers the results, and maps each failure to the formal problem representation during diagnosis.

Failure taxonomy classification in formal robotics research is the process of categorizing experiment failures into perception, planning, execution, physics, or spec categories. It links these failures to formal problem representations to guide subsequent challenge or formalize steps.

You do not strictly need wandb to categorize minimal system failures. The diagnosis process can gather results from wandb APIs or local logs, classifying the failures into taxonomy categories and mapping them to the formal problem representation.