Camera-ready submission reviewer suggestions

[afr2903]

• Clarify Table 3 Sample Counts: Add counts for standard and enriched datasets
• Specify Table 4 Sample Counts for Categories A, B, C: Insert actual sample numbers
• Provide Example for Category A (Navigation & Following): Extract or create a sample command
• Provide Example for Category B (Object Manipulation): Extract or create a sample command
• Provide Example for Category C (Interaction & Querying): Extract or create a sample command
• Expand on Table 4 Category Descriptions: Add 1-2 lines explaining each type
• Add a section distinguishing out-of-domain entity handling: Summarize limitations and possible mitigations
• Explicitly mention current limitation for out-of-distribution/non-existent entities: Future work
• Add discussion on expanding dataset via user feedback: Future work
• Explicitly acknowledge RoboCup GPSR context limitation
• Mention multi-modal grounding as future direction
• Add possibility of Chain-of-Thought prompting as future work
• Add example command(s) in different languages? At least an example
• Explicitly state model limitation to English: mention dataset is english because... (make credible justification)
• Enhance explanation of Schema-Aligned Parsing (SAP)
• Add more examples of hallucinated vs. real actions
• Tighten section headings and transitions for clarity (reviewer 3 about hard to follow)
• Proofread for overall clarity; minimize jargon