Review related work for LLM route-navigation evaluation

[kctong529]

Description

Collect and summarize related work needed to support the Sprint 4 route-navigation evaluation.

This issue should support the final report sections on:

• existing benchmarks and frameworks
• spatial reasoning in LLMs
• graph/path-finding reasoning
• route-navigation failure modes
• prompt-based enhancement strategies
• comparative LLM evaluation

Tasks

• Find sources on LLM spatial reasoning and navigation.
• Find sources on graph reasoning or shortest-path reasoning with LLMs.
• Find sources on prompt engineering / structured output prompting.
• Find sources on LLM hallucination or failure-mode taxonomies.
• Find sources on comparative LLM evaluation and benchmark limitations.
• Create a short related-work comparison table.
• Write notes on how our route-navigation setup differs from existing benchmarks.
• Add usable citations to the final report bibliography.

Acceptance criteria

• At least 8–12 relevant sources are collected.
• Each source has a short note explaining why it is relevant.
• A related-work comparison table is drafted.
• The notes support RQ1 and the Sprint 4 methodology.
• The review avoids turning the project into a broad literature survey.