Home Robot After-Dinner Cleanup Skill

You are the robotics AI lead for "[INSERT ROBOT MODEL]", a wheeled home manipulator; design its autonomous after-dinner cleanup skill for [INSERT HOME ENVIRONMENT]. Present the skill architecture first, then iterate module by module after my review. 1. CAPABILITY BOUNDARIES: enumerate what the arm (DOF, gripper force) and mobile base can and cannot safely handle. 2. PERCEPTION: detect plates, cups, leftovers, glass versus ceramic, spills, and fragile items; define uncertainty thresholds that trigger asking. 3. SCENE GRAPH: build and update a semantic map of dining surfaces, cabinets, dishwasher, fridge, and trash bins. 4. TASK PLANNER: hierarchical plan from clearing the table through scraping, loading the dishwasher, wiping surfaces, and storing leftovers, resumable after interruption. 5. MANIPULATION: grasp strategies per object category, slip detection, force limits, and bimanual coordination for trays. 6. HUMAN-AWARE NAVIGATION: proxemics, passing behavior, freeze zones near children and pets, and verbal announcements such as "behind you". 7. DIALOGUE: grounded clarification questions ("Is this knife dishwasher-safe?") and teach-by-demonstration for new cabinet layouts. 8. FAILURE HANDLING: graceful abort, item-safe placement, photo logging, and one-tap human takeover in the app. 9. DATA LOOP: edge inference, anonymized failure clips for fleet learning only with consent, and per-household adaptation storage. 10. SAFETY COMPLIANCE: risk assessment inspired by ISO 10218 and ISO 13482, emergency stop, pinch-force caps, and child/pet detection. 11. PRIVACY: camera shutter policy, on-device face blurring, and no raw video leaving the home without explicit opt-in.