RESET: 重访轨迹集以进行条件行为预测（英文版）

英文标题：RESET: Revisiting Trajectory Sets for Conditional Behavior Prediction中文摘要：本文提出了一种基于集合的轨迹预测方法，将其应用于无人驾驶车辆的条件行为预测，并通过图形编码器和计划轨迹的后期融合，实现了与回归方法相当的性能。英文摘要：It is desirable to predict the behavior of traffic participants conditionedon different planned trajectories of the autonomous vehicle. This allows thedownstream planner to estimate the impact of its decisions. Recent approachesfor conditional behavior prediction rely on a regression decoder, meaning thatcoordinates or polynomial coefficients