The independent variable robot releases the cross-tool body motion analyzer X-Tokenizer.

On June 30th, the autonomous robot released the X-Tokenizer, a cross-positional motion parser. According to the introduction, X-Tokenizer adopts an "encoder-semantic residual quantization-decoder" architecture, using SRQ to replace traditional standard residual vector quantization. This layer separates the action intent and has been pre-trained on 2.4 million trajectories covering 17 series of mechanical arms.