Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware
Paper • 2304.13705 • Published • 9
ACT: Optimized for Qualcomm Devices
ACT (Action Chunking with Transformers) is a robotic policy model that is trained to predict the next chunk of actions that the robotic hand is expected to perform.
This is based on the implementation of ACT found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.
Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.
Getting Started
There are two ways to deploy this model on your device:
Option 1: Download Pre-Exported Models
Below are pre-exported model assets ready for deployment.
| Runtime | Precision | Chipset | SDK Versions | Download |
|---|---|---|---|---|
| ONNX | float | Universal | QAIRT 2.50, ONNX Runtime 1.27.1 | Download |
| QNN_DLC | float | Universal | QAIRT 2.50 | Download |
| TFLITE | float | Universal | QAIRT 2.50 | Download |
For more device-specific assets and performance metrics, visit ACT on Qualcomm® AI Hub.
Option 2: Export with Custom Configurations
Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:
- Custom weights (e.g., fine-tuned checkpoints)
- Custom input shapes
- Target device and runtime configurations
This option is ideal if you need to customize the model beyond the default configuration provided here.
See our repository for ACT on GitHub for usage instructions.
Model Details
Model Type: Model_use_case.robotics
Model Stats:
- Input resolution: 480x640
- Model checkpoint: act
- Model size (float): 255M
- Number of parameters: 83.92M
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| ACT | ONNX | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 5.917 ms | 1 - 359 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 6.714 ms | 2 - 373 MB | NPU |
| ACT | ONNX | float | Snapdragon® X2 Elite | 6.62 ms | 4 - 4 MB | NPU |
| ACT | ONNX | float | Snapdragon® X Elite | 12.571 ms | 64 - 64 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 8.293 ms | 0 - 465 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 17.877 ms | 0 - 310 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 16.888 ms | 3 - 11 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 11.821 ms | 0 - 78 MB | NPU |
| ACT | ONNX | float | Qualcomm® QCS8450 | 17.877 ms | 0 - 310 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 17.657 ms | 3 - 10 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 12.571 ms | 64 - 64 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 6.714 ms | 2 - 373 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 4.301 ms | 4 - 381 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 5.271 ms | 0 - 284 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® X2 Elite | 5.26 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® X Elite | 11.174 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 6.998 ms | 0 - 479 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 15.221 ms | 4 - 251 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 14.698 ms | 4 - 10 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 10.368 ms | 4 - 6 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8650P | 14.499 ms | 1 - 271 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8255P | 14.499 ms | 1 - 271 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8450 | 15.221 ms | 4 - 251 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 14.771 ms | 4 - 9 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 11.174 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 5.271 ms | 0 - 284 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8295P | 16.621 ms | 0 - 200 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Elite Gen 5 For Galaxy Mobile | 3.838 ms | 0 - 376 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.757 ms | 0 - 323 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 6.162 ms | 0 - 469 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 13.527 ms | 0 - 264 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ IQ-8275 | 14.362 ms | 0 - 72 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.635 ms | 0 - 7 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8650P | 13.799 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8255P | 13.799 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8450 | 13.527 ms | 0 - 264 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 14.033 ms | 0 - 71 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.757 ms | 0 - 323 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8295P | 15.651 ms | 0 - 198 MB | NPU |
License
- The license for the original implementation of ACT can be found here.
References
Community
- Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
- For questions or feedback please reach out to us.