Use when writing or revising Chapter 2 literature-review prose from structured paper summaries in papers/download or papers/relevant, or when a chapter section needs to be rewritten to argue a gap rather than list papers.
Read chapters/AGENTS.md and papers/AGENTS.md before writing. Chapter 2 constraints (no subsections, no gap codes in prose, tables integrated with motivation and interpretation) are fully specified there.
Every substantial paragraph in Chapter 2 should follow:
The last sentence of a paragraph should always be the argument, not a citation.
Keep this visible across the chapter:
| Stage | Content |
|---|---|
| 1 | Fixed-base limitation → why VMS are needed |
| 2 | Model-free / data-driven approaches → why model-based preferred |
| 3 | Non-geometric model-based → coordinate limitations |
| 4 | Geometric model-based → manifold advantages |
| 5 | Energy-based modelling → chosen cross-cutting path |
| 6 | Port-Hamiltonian interaction control → chosen controller framing |
| 7 | Safe interaction → the harder challenge, Aim 3 |
Add a table only when it makes a comparison, chronology, or classification clearer than prose. Every table must:
A table that appears without motivation or interpretation violates supervisor constraints.
| Pattern | Why to avoid |
|---|---|
| Paragraph that only lists papers with brief descriptions | Reads as a catalogue, not an argument |
| Gap stated without linking it to the thesis aims | The reader cannot see why the gap matters |
| Table without motivation or interpretation | Supervisor constraint; also leaves the reader to guess the point |
| Subsection heading in Chapter 2 | Supervisor constraint |
| Gap codes (M1, C2, H3) in thesis prose | Supervisor constraint |
| Section ending on a citation | Section conclusions should be synthetic statements, not references |
| Forcing a weak paper into a strong claim | Weakens the argument; acknowledge weak relevance plainly |