Hi：
Thank you for your excellent work！
I have a slight confusion about the definition of the Multi-World Prediction metric.
Does it refer to:
（1）In one scenario, there are M worlds and A agents. In each world, each agent predicts K trajectories, resulting in a total of M * A * K trajectories. Each world calculates an avgMinFDE, and the world with the smallest avgMinFDE is considered the best world, and the final avgMinFDE of this senario is the avgMinFDE of the best world.
Or:
（2）In one scenario, there are M worlds and A agents. In each world, each agent predicts 1 trajectory, resulting in a total of M * A trajectories. Each world calculates an avgFDE, and the world with the smallest avgFDE is considered the best world, with this value taken as the avgMinFDE.
Thanks!
Hi：
Thank you for your excellent work！
I have a slight confusion about the definition of the Multi-World Prediction metric.
Does it refer to:
（1）In one scenario, there are M worlds and A agents. In each world, each agent predicts K trajectories, resulting in a total of M * A * K trajectories. Each world calculates an avgMinFDE, and the world with the smallest avgMinFDE is considered the best world, and the final avgMinFDE of this senario is the avgMinFDE of the best world.
Or:
（2）In one scenario, there are M worlds and A agents. In each world, each agent predicts 1 trajectory, resulting in a total of M * A trajectories. Each world calculates an avgFDE, and the world with the smallest avgFDE is considered the best world, with this value taken as the avgMinFDE.
Thanks!