tower-benchmarkcc-by-4.0

Tower-SimData

A simulated demonstration dataset for bimanual Jenga manipulation research containing robot states, commands, end-effector poses, and four camera streams organized across three task types with 2-3 layer variants.

Downloads61
Episodes331

Why This Matters for Physical AI

This dataset enables research on bimanual robotic manipulation and imitation learning for complex structured object handling tasks through high-fidelity simulation data with multi-view observations and precise state-action trajectories.

Technical Profile

Modalities
rgbproprioception
Robot Embodiments
bimanual manipulator
Action Space
joint_positions
Environment
simulation
Task Types
manipulationtower_transfercross_tower_transfervertical_stacking
Episodes
331
Data Format
HDF5
Annotation Types
action_labelstask_success_labels
License
cc-by-4.0
Part of the TOWER family

Access

Need custom rgb data?

Claru builds purpose-built datasets for simulation applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets