Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware
Paper • 2304.13705 • Published • 7
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 2.45, ONNX Runtime 1.25.0 | Download |
| QNN_DLC | float | Universal | QAIRT 2.45 | Download |
| TFLITE | float | Universal | QAIRT 2.45 | 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:
- Model checkpoint: act
- Input resolution: 480x640
- Number of parameters: 83.92M
- Model size (float): 255M
Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |
|---|---|---|---|---|---|---|
| ACT | ONNX | float | Snapdragon® X2 Elite | 6.716 ms | 4 - 4 MB | NPU |
| ACT | ONNX | float | Snapdragon® X Elite | 12.675 ms | 63 - 63 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 8.627 ms | 4 - 511 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 19.486 ms | 4 - 320 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 12.151 ms | 0 - 79 MB | NPU |
| ACT | ONNX | float | Qualcomm® QCS8450 | 19.486 ms | 4 - 320 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 18.022 ms | 4 - 10 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 5.984 ms | 1 - 444 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Elite Mobile | 6.868 ms | 0 - 408 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 6.868 ms | 0 - 408 MB | NPU |
| ACT | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 12.675 ms | 63 - 63 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® X2 Elite | 4.804 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® X Elite | 9.198 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 6.074 ms | 0 - 336 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 16.419 ms | 3 - 253 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8275 | 44.71 ms | 1 - 285 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.531 ms | 4 - 5 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8775P | 13.792 ms | 1 - 269 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8650P | 13.792 ms | 1 - 269 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8255P | 13.792 ms | 1 - 269 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8450 | 16.419 ms | 3 - 253 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 13.728 ms | 4 - 9 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.745 ms | 4 - 300 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA7255P | 44.71 ms | 1 - 285 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 4.717 ms | 4 - 273 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8295P | 15.254 ms | 0 - 204 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 4.717 ms | 4 - 273 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 9.198 ms | 4 - 4 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 6.105 ms | 0 - 409 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 16.37 ms | 0 - 271 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8275 | 44.735 ms | 0 - 326 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.448 ms | 0 - 3 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8775P | 13.83 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8650P | 13.83 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8255P | 13.83 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8450 | 16.37 ms | 0 - 271 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 13.582 ms | 0 - 71 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.791 ms | 0 - 324 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA7255P | 44.735 ms | 0 - 326 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Elite Mobile | 4.733 ms | 0 - 283 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8295P | 15.739 ms | 0 - 208 MB | NPU |
| ACT | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.733 ms | 0 - 283 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.