Using a multilayer perciptron (MLP) in combination with the first cleanup method to simulate the trajectories of five different objects within a single SSP. The MLP takes in the last estimate of SSP and the object's velocities represented as a single SSP as input and estimates the SSP at the next time step. Starting with the initial SSP, at each 0.05 s time step, the SSP is updated using the MLP and then “cleaned up” to simulate the objects' motion. The dashed lines represent the object trajectories decoded from this simulation, while the solid lines represent the ground-truth object trajectories.
This site uses cookies. By continuing to use our website, you are agreeing to our privacy policy. No content on this site may be used to train artificial intelligence systems without permission in writing from the MIT Press.