Benchmark RadarUESF-Bench
UESF-Bench evaluates embodied agents on unified language-guided human seeking and following in dynamic environments, covering semantic-guided exploration, behavior switching, and identity grounding across single- and multi-person settings. Scoring uses success metrics for both phases.
- Released
- 2026-07-15
- Readiness
- Paper only
- Primary field
- Robotics & Autonomous Systems
Why it matters
Existing benchmarks assume the target is visible at start, missing realistic scenarios where agents must first find and then follow. UESF-Bench provides a unified evaluation to advance embodied agents in more practical human-robot interaction.
Motivation
Language-guided human following is an important capability for embodied agents, but existing benchmarks typically assume that the target person is visible at the start of an episode.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.