How to generate multimodal trajectories using flow planner?

[CuberrChen]

Thanks your great work. I’m wondering whether flow planner is a single modality trajectory planning framework? How to extend it to output multimodal trajectories?

[ChuXX325]

可以多模态，多采样几个噪声就行了，但是模型不会给多模态轨迹打分，还需要一个打分模型或者打分规则