1 paper
Chushan Zhang, Ruihan Lu, Jinguang Tong +3
Chunked vision-language-action (VLA) policies predict multi-step robot controls, conditioning each update on the current visual observation alone. Yet robot actions cause contact,…