Recordings of a wooden cube repeatedly dropped by a Kinova Gen3 Robot Arm. The dataset includes audio from a 7-channel microphone array and video from wrist and ceiling cameras, capturing the cube's impact sound and trajectory. It was created by Fanjun Bu of Cornell University to address robot error recovery based on object permanence.
Use Cases
- Training multimodal models for sound localization based on impact audio recordings
- Developing object permanence algorithms for robots based on partial and complete trajectory data
- Benchmarking robot error recovery strategies using synchronized audio and visual data
- Studying the physics of bouncing objects from audio-visual sensor fusion
Strengths
- Data includes synchronized audio (WAV files) and visual recordings from multiple camera perspectives
- Contains two versions of trajectory data: partial from a wrist camera and complete from merged trajectories
- Focuses on a specific, controlled scenario of a 3cm cube dropped from a 0.3-meter height
Limitations
- Row count and dataset size are unknown, which may limit suitability assessment
- Column-level documentation is absent; field semantics must be inferred after download
- Last update date is unknown; freshness unverified
Provenance
- Source
- Cornell University
- Collection Method
- Collected by a Kinova Gen3 Robot Arm repeatedly picking up and dropping a wooden cube, recorded with a microphone array and cameras.
- Time Range
- null
- Freshness
- null
- Geography
- null