Datasheet-grounded review of KiCad schematics

The research direction

In my master’s thesis I am investigating the use of a local, datasheet-assisted approach to reviewing component-specific faults in KiCad schematics beyond conventional electrical rule checking.

The work will connect the structure of a schematic with manufacturer datasheet. I will develop the local system and compare the review approaches.

The review workflow

I will design a workflow that combines four elements:

• Component–pin–net relationships will represent how the parts of a schematic are connected.
• Manufacturer datasheet evidence will inform component-specific review.
• Deterministic checks will support rule-based review.
• A local language model will form part of the review approach.

I will aim to review schematic faults with evidence tied to the relevant component and connection.

How it will be evaluated

I will define an evaluation protocol comparing rule-based, model-based, retrieval-assisted and hybrid approaches. The comparison will examine several aspects of review quality:

• Fault detection: whether an approach identifies the faults under evaluation.
• Localization: whether a reported issue points to the relevant part of the schematic.
• False warnings: whether an approach reports issues that are not faults.
• Evidence quality: whether the review provides suitable supporting evidence.

Planned stages

I will define the review workflow and evaluation protocol, develop the local system and carry out the comparison during my thesis, planned for September 2026 to January 2027.