Grasping-Dataset: Robotic Grasp Stability Predictions for a Ball

A dataset for investigating machine learning predictions of robotic grasp stability, created by Shadow Robot as part of their Smart Grasping System development. It contains annotated grasp quality scores and data gathered from robot joints, including position, velocity, and effort. The dataset is currently restricted to grasping a ball due to simulation limitations.

Use Cases

• Predicting grasp stability based on joint position, velocity, and effort data.
• Training machine learning models for robotic manipulation tasks.
• Benchmarking algorithms for grasp quality assessment.
• Investigating the relationship between joint sensor data and grasp success.

Strengths

• Dataset is annotated with an objective grasp quality metric.
• Contains multiple data streams from robot joints, including position, velocity, and effort.

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.
• Scope is restricted to grasping a single object (a ball) due to simulation constraints.

Provenance

Source

Shadow Robot

Collection Method

Created as part of Smart Grasping System development, likely from simulation or robotic hardware.