HygieneRoboBench: Benchmarking Hygiene-Aware Planning for Household Robots
The benchmark tests whether household robot planners can keep tasks safe after contamination events change the plan.
HygieneRoboBench includes 624 instances across 134 household task families. It tracks contamination through two grippers and shared objects, along with treatment costs and user priorities. The paper reports that LLM-based and symbolic planners can finish tasks safely without finding the lowest-cost safe plan. Its Hygiene-NSP system combines LLM grounding, contact-history reconstruction, and CP-SAT, reaching 94.4% safe resolution and 90.4% optimal safe resolution on the full dataset. ArXiv · AI/CL/LG's note
HygieneRoboBench includes 624 instances across 134 household task families. It tracks contamination through two grippers and shared objects, along with treatment costs and user priorities. The paper reports that LLM-based and symbolic planners can finish tasks safely without finding the lowest-cost safe plan. Its Hygiene-NSP system combines LLM grounding, contact-history reconstruction, and CP-SAT, reaching 94.4% safe resolution and 90.4% optimal safe resolution on the full dataset. ArXiv · AI/CL/LG's note
score 4