Collision Indeterminacy Prediction via Stochastic Trajectory Generation


At a glance

Trajectory prediction is an important component of collision avoidance systems for autonomous vehicles. The goal is to predict potential collisions between traffic participants or obstacles on the road and to take appropriate actions to avoid or mitigate them. Collisions often occur in infrastructures where the paths of multiple participants frequently intersect, such as intersections, ramps, and weaving areas.
In this project, we propose a stochastic trajectory generation to determine the possible collision between traffic participants by taking into account the uncertainty and variability of the future movements of the ego vehicle and other participants.
Stochastic trajectory generation provides a more informative representation of the uncertainty and variability in object movement, as the motion of the object is represented as a probability distribution over possible future states.

Principal Investigatorsresearchersthemes

Maria Laura Delle Monache

Mark Hansen

Han Wang

Stochastic trajectory generation, Generative adversarial networks, 3D Simulation