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 2.42, 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® 8 Elite Gen 5 Mobile | 5.851 ms | 4 - 426 MB | NPU |
| ACT | ONNX | float | Snapdragon® X2 Elite | 6.772 ms | 209 - 209 MB | NPU |
| ACT | ONNX | float | Snapdragon® X Elite | 12.196 ms | 177 - 177 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 8.49 ms | 4 - 499 MB | NPU |
| ACT | ONNX | float | Qualcomm® QCS8550 (Proxy) | 11.546 ms | 4 - 124 MB | NPU |
| ACT | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 6.822 ms | 0 - 422 MB | NPU |
| ACT | ONNX | float | Qualcomm® QCS9075 | 19.721 ms | 4 - 49 MB | NPU |
| ACT | ONNX | float | Qualcomm® QCS8750 | 6.822 ms | 0 - 422 MB | NPU |
| ACT | ONNX | float | Qualcomm® QCS7181 | 12.196 ms | 177 - 177 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.732 ms | 3 - 294 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® X2 Elite | 4.835 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® X Elite | 9.215 ms | 4 - 4 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 6.058 ms | 4 - 341 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8275 | 44.729 ms | 1 - 286 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 8.582 ms | 4 - 6 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8775P | 13.806 ms | 0 - 268 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8650P | 13.806 ms | 0 - 268 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8255P | 13.806 ms | 0 - 268 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 16.484 ms | 2 - 250 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA7255P | 44.729 ms | 1 - 286 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® SA8295P | 15.24 ms | 0 - 204 MB | NPU |
| ACT | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.698 ms | 4 - 274 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS9075 | 15.182 ms | 4 - 9 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS8750 | 4.698 ms | 4 - 274 MB | NPU |
| ACT | QNN_DLC | float | Qualcomm® QCS7181 | 9.215 ms | 4 - 4 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.736 ms | 0 - 319 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 6.109 ms | 0 - 411 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8275 | 44.754 ms | 0 - 325 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 8.402 ms | 0 - 3 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8775P | 13.782 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8650P | 13.782 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8255P | 13.782 ms | 0 - 279 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 16.426 ms | 0 - 269 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA7255P | 44.754 ms | 0 - 325 MB | NPU |
| ACT | TFLITE | float | Qualcomm® SA8295P | 15.732 ms | 0 - 207 MB | NPU |
| ACT | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 4.733 ms | 0 - 282 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS9075 | 15.187 ms | 0 - 71 MB | NPU |
| ACT | TFLITE | float | Qualcomm® QCS8750 | 4.733 ms | 0 - 282 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.