PrivateMap-Bench includes 15 real robot runs across 3 layout variants, producing 300 filtered artifacts, 2310 privacy rows, 5580 utility rows, and 1126 aggregate rows. It is a benchmark for studying the privacy–utility trade-off in indoor SLAM map sharing, evaluating floor-plan recovery, trajectory leakage, and navigation utility. The dataset was created by an anonymous author and last updated on May 7, 2026.
Use Cases
- Benchmark privacy leakage of filtered robot maps based on floor-plan recovery metrics
- Evaluate navigation utility of anonymized trajectories based on A* navigation utility scores
- Study the trade-off between information recoverability and task utility in map sharing
- Audit routine leakage from robot paths based on trajectory similarity metrics
Strengths
- Includes 15 real robot runs across 3 layout variants
- Provides 300 filtered artifacts and over 9000 rows of privacy and utility measurements
- Evaluates multiple specific metrics, including occupied-cell F1 scores and visit-order similarity
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Raw ROS bags and unreviewed images are withheld by default
- Row count for the main data tables is unknown, which may limit suitability assessment
Provenance
- Source
- Benchmark
- Collection Method
- Real robot runs producing filtered maps and trajectories
- Freshness
- Last updated 2026-05-07 10:04:17