PDMS
A composite scoring metric in the NAVSIM benchmark for measuring autonomous driving planning quality. It is formed by a weighted combination of sub-metrics including collision avoidance, drivable-area compliance, comfort and progress; higher values indicate safer planning that more closely resembles human driving behavior.
Timeline (last 90 days)
On the navtest benchmark of NAVSIM v1, VITAL reached 90.8 PDMS with just one action denoising step, only 0.13 points behind the score of ten-step inference, with a 3.29x improvement in end-to-end inference speed.
PDMS (Predictive Driver Model Score) is a composite scoring metric used in the NAVSIM benchmark to measure planning quality, covering dimensions such as collision avoidance, drivable-area compliance, comfort and progress, where a higher value indicates safer planning that is closer to human driving behavior.
All Facts (2)
On the navtest benchmark of NAVSIM v1, VITAL reached 90.8 PDMS with just one action denoising step, only 0.13 points behind the score of ten-step inference, with a 3.29x improvement in end-to-end inference speed.
50%UnverifiedPDMS (Predictive Driver Model Score) is a composite scoring metric used in the NAVSIM benchmark to measure planning quality, covering dimensions such as collision avoidance, drivable-area compliance, comfort and progress, where a higher value indicates safer planning that is closer to human driving behavior.
50%