metadata
license: mit
tags:
- robotics
- act
- vla
- gearbox-assembly
- imitation-learning
metrics:
- l1_loss
- kl_divergence
ACT Policy for Gearbox Assembly (Filtered Demos)
This model is an Action Chunking with Transformers (ACT) policy trained to perform gearbox assembly tasks. It was trained using behavior cloning on a combination of rocochallenge2025 and additional collected datasets.
Model Details
- Model Type: ACT (Action Chunking with Transformers)
- Policy Class: ACT
- Backbone: ResNet-18
- Training Dataset: Integrated dataset (
rocochallenge2025+temp_new_dataset) containing 241 episodes. - Episode Length: Fixed to 12,600 steps (padded/truncated) to handle variable length recordings.
Training Configuration
- Task Name:
sim_gearbox_assembly_demos_filtered - Batch Size: 32
- Chunk Size (Action Horizon): 100
- KL Weight: 10
- Hidden Dimension: 512
- Feedforward Dimension: 3200
- Learning Rate: 1e-5
- Num Epochs: ~9500 (Early stopped/Interrupted)
- Seed: 0
Inputs and Outputs
- Observations:
head_rgb(240x320)left_hand_rgb(240x320)right_hand_rgb(240x320)qpos(14-dim joint positions)
- Actions:
- 14-dim combined action vector (7-dim left arm + 7-dim right arm)
Usages
This model can be loaded using the ACTPolicy class. Ensure dataset_stats.pkl is loaded to normalize/unnormalize observations and actions correctly.
from policy import ACTPolicy
import pickle
# Load stats
with open('dataset_stats.pkl', 'rb') as f:
stats = pickle.load(f)
# Load policy
policy = ACTPolicy(config)
policy.load_state_dict(torch.load('policy_best.ckpt'))
policy.cuda()
policy.eval()