SMILE

We compare SMILE with raw-action SmolVLA to test whether predicting B-spline coefficients improves real-robot task success at matched prediction and execution horizons. No official code is released, so we reimplemented SMILE’s action interface from the paper on SmolVLA’s pretrained backbone. The action expert denoises 8 cubic B-spline control points that are decoded into the action chunk. Both models predict relative translation and rotation-vector actions with absolute gripper widths and use matched training settings within each task.

• Prediction length. Both models predict 50 action steps, following the SmolVLA recipe. This prediction length is not confirmed in the SMILE paper.
• Synchronous execution. Both models execute 15 actions per plan, 0.5 seconds at 30 Hz, and the station waits for each plan to finish before requesting the next.
• Episode-end supervision. We fit splines only to valid samples and supervise only the coefficient directions they determine, filling unconstrained directions with the dataset mean. This retains partial windows in both arms; full windows use the paper’s coefficient-space loss.
• Fine-tuning budget. Both arms use batch 64, learning rate 1e-4 and at least 20k steps or three epochs, whichever is greater, without image augmentation or a validation split. Exact parity cannot be established because the paper does not report a complete training recipe.