Xarm Push Medium Image: Robot Manipulation Training Data

800 episodes of robot manipulation data created using the LeRobot framework. The dataset contains 20,000 frames recorded at 15 frames per second. It is structured for training imitation learning models on a single push task.

Use Cases

• Train robot control policies based on visual observations
• Benchmark imitation learning algorithms on a defined push task
• Analyze robot state-action sequences for a single manipulation objective

Strengths

• 800 episodes provide a substantial number of training trajectories
• 20,000 frames offer a significant volume of sequential visual data
• Data is structured into a single chunk for straightforward access

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

lerobot

Collection Method

Created using the LeRobot framework.

Freshness

Last updated 2026-06-08 18:23:21; freshness should be verified