SlackDrive: Reclaiming Runtime Slack for Adaptive Driving Inference
Driving world-action models improve planning by coupling multimodal reasoning with future prediction, but their growing inference cost increasingly conflicts with the real-time latency requirements of vehicle control. Existing acceleration methods reduce tokens, layers, or sampling steps with policies selected prior to...
Xiao-Huan Pei, Heng-Guang Zhou, Yuan-Hao Ban et al.
· 0 citations