[Lane Change] a combined model and learning based framework for interaction aware maneuver prediction, IEEE Trans. on IV, 2016

Summary

. Predict future intentions of the vehicles

. First predict future intentions ({LLC, RLC, LK}) by using model-based algorithm

. Second, predict maneuvers (lateral={LLC, RLC, LK}, longitunial={accelerations}) from the past trajectories and the future intentions, statistically

 

 

1. Input

. position, acceleration, dist to the next highway junction, type of lane marking, distance to lane end, current speed limit

 

2. Algorithm

1) Interaction-aware intention prediction

 . predict the future intentions of the vehicles based on the cost optimization approach

 . Inputs are the all of aboves

 . Output is a series of maneuvers, {LLC, RLC, LK} and estimated multi-variate Gaussian parameters

 

 

 

2) Lateral and Longitunial maneuver prediction

 . Using the outputs from 2-1,

   a) lateral maneuver prediction : Baysian filter

   b) longitunial accelerations : mixtured Gaussian

 

 

 

 

3. Results