Abstract

A technique is proposed herein for adding a learned time-budget output to a robotic Vision-Language-Action (VLA) policy model. Along with each robot action, the model emits an action-specific temporal parameter indicating how long the action is expected to take or what deadline applies. Inference schedulers, edge/cloud placement systems, and network controllers use the temporal parameter to determine how to process the next action before the current action completes.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS