Task-oriented Tool Manipulation with Robotic Dexterous Hands: A Knowledge Graph Approach from Fingers to Functionality

This is the F2F Knowledge Graph data set and code of paper Task-oriented Tool Manipulation with Robotic Dexterous Hands: A Knowledge Graph Approach from Fingers to Functionality.

Dependencies

• Python 3.6+
• PyTorch 1.0+

Running the code

Usage

bash runs.sh {train | valid | test} {HAKE} {F2F-V2 | your Knowledge Graph} <gpu_id> \
<save_id> <train_batch_size> <negative_sample_size> <hidden_dim> <gamma> <alpha> \
<learning_rate> <num_train_steps> <test_batch_size> [modulus_weight] [phase_weight]

• { | }: Mutually exclusive items. Choose one from them.
• < >: Placeholder for which you must supply a value.
• [ ]: Optional items.

Remark: [modulus_weight] and [phase_weight] are available only for the HAKE model.

To reproduce the results of HAKE, run the following commands.

HAKE

bash runs.sh train HAKE F2F-V2 0 3 512 1024 500 12 1 0.00005 20000 16 1 1

Citation

If you find this code useful, please consider citing the following paper.

Acknowledgement

We refer to the code of HAKE. Thanks for their contributions.