[SIGGRAPH Asia 2025] Physics-Based Motion Imitation with Adversarial Differential Discriminators скачать в хорошем качестве

[SIGGRAPH Asia 2025] Physics-Based Motion Imitation with Adversarial Differential Discriminators

Tired of designing and tuning reward functions? ADD can help! ADD is an adversarial Multi-Objective Optimization (MOO) solution that is broadly applicable to a range of problems, including motion tracking. It automatically balances the various objectives, reducing the burden of manual reward engineering and weight tuning. Website: https://add-moo.github.io/ Arxiv: https://arxiv.org/abs/2505.04961 Code: https://github.com/xbpeng/MimicKit Timeline: 00:00 Overview 01:09 ADD for Motion Imitation 02:15 Humanoid Motion Imitation 02:29 Humanoid (Parkour) 02:45 Humanoid (Datasets) 03:02 Performance Comparison 03:41 Robot Motion Imitation 03:54 Non-motion-imitation tasks 04:09 Quadruped Locomotion 04:24 Walker Benchmark 04:55 Ablation