ACT SO-101 Pick-and-Place
A task-specific ACT policy trained to place a grey eraser in a white basket, compiled for the SiMa.ai Modalix DevKit.
| Property | Value |
|---|---|
| Training checkpoint | Step 100,000 |
| Compiled output | 100 actions × 6 joints |
| Precision | BF16 |
| Robot | SO-101 follower |
| Task | Place the grey eraser in the white basket |
| Bundle ID | act-so101-grey-eraser-pickplace-100k-bf16-1879ca7c |
This repository contains the deployable compiled bundle, reference fixtures and SO-101 runtime assets. It does not contain the training checkpoint or dataset. The policy is specific to the reference task and workspace.
Prerequisites
- Modalix DevKit with NVMe storage and a compatible SiMa.ai runtime.
- PoLiMa v0.3 installed from its source checkout on the deployment host and DevKit.
- A Linux deployment host, or WSL on Windows, with SSH, rsync and the Hugging Face CLI.
- SO-101 follower with six STS3215 motors (IDs 1–6), securely mounted and powered.
- Two USB cameras connected to the DevKit and framed like the training setup.
The release checks used ModelSDK 2.1.3, Neat 0.5.0 and sima_lmm 0.5.0.
Download and deploy
hf download simaai/ACT-SO101-PickPlace --local-dir ACT-SO101-PickPlace
cd ACT-SO101-PickPlace
sha256sum --check SHA256SUMS
read -r -p "DevKit login (username@address): " devkit_login
polima deploy --bundle ./bundle --devkit "$devkit_login" --start
Deployment installs and activates the bundle and starts the inference server. It does not start robot motion.
Check inference
polima run --bundle ./bundle --devkit "$devkit_login" --fixture --repeat 20
This replays the packaged observations without cameras or robot motion and checks numerical agreement. It does not measure physical task success.
Run the robot
Use Studio to inspect both camera views and confirm the selected arm, cameras and task. Calibrate each physical follower arm before its first run.
polima studio start
polima robot calibrate
polima robot run
Reference cameras are a Logitech C920 overhead camera and a Sonix USB2.0 CAM1 wrist camera, capturing 640 × 480 MJPG at 30 frames/s. The overhead lens was 49 cm above the table; the basket was approximately 32 cm from the arm end. The basket measured 17.5 × 12 × 6 cm and the eraser 12.5 × 5 × 2 cm.
Training data
Trained on 769 demonstrations from SO-101 Grey Eraser Pick-and-Place. The full collection contains 961 episodes; 192 episodes were held out.
Validation
On the reference Modalix runtime, 30 measured requests after five warm-ups gave 13.23 ms p50, 13.39 ms p95 and 13.45 ms p99 model-server latency. The packaged fixture passed compiled-device fidelity checks. Sixteen held-out frames passed cosine similarity ≥ 0.999 and normalized mean absolute error ≤ 0.01 against PyTorch. Physical operation was verified on the reference SO-101 setup; a controlled success rate was not recorded.
See the PoLiMa benchmark documentation for methods, stage timings, memory, disk use and limitations.
Package contents
bundle/ contains six compiled ELFs, the execution plan, normalization data,
reference observations and SO-101 launch assets. SHA256SUMS covers every
supplied file except the checksum file itself.
Limitations
The policy is specific to the reference robot, object, basket and camera views. Different geometry, framing, lighting or calibration can change behavior. Text changes do not teach ACT a new behavior at runtime.
License
The package is released under Apache-2.0. See LICENSE.