Get Zlio on your home screen
One tap, full screen, like an app.
A research paper presents PDRVLA, a vision-language-action robot model whose lightweight Physical Dynamics Module extracts physical tokens from vision instead of generating future video, cutting latency in the control loop.