Search papers, labs, and topics across Lattice.
3
0
5
Robots can now learn dexterous manipulation skills across different hand designs, thanks to a new Transformer architecture that treats actions as a flexible arrangement of joint movements, rather than a fixed sequence.
Achieve significant inference speedups in visual autoregressive models without retraining by pruning redundant tokens based on a novel structure-texture importance criterion.
By decoupling coarse action consistency from fine-grained variations, PF-DAG achieves state-of-the-art imitation learning performance in robotic manipulation tasks.
