Learning a universal policy across different robot morphologies can significantly improve learning efficiency and enable zero-shot generalization to unseen morphologies. However, learning a highly performant universal policy requires sophisticated architectures like transformers (TF) that have larger memory and computational cost than simpler multi-layer perceptrons (MLP). To achieve both good performance like TF and high efficiency like MLP at inference time, we propose HyperDistill, which consists of: (1) A morphology-conditioned hypernetwork (HN) that generates robot-wise MLP policies, and (2) A policy distillation approach that is essential for successful training. We show that on UNIMAL, a benchmark with hundreds of diverse morphologies, HyperDistill performs as well as a universal TF teacher policy on both training and unseen test robots, but reduces model size by 6-14 times, and computational cost by 67-160 times in different environments. Our analysis attributes the efficiency advantage of HyperDistill at inference time to knowledge decoupling, i.e., the ability to decouple inter-task and intra-task knowledge, a general principle that could also be applied to improve inference efficiency in other domains.

翻译：学习跨不同机器人形态的通用策略可显著提升学习效率，并实现对新形态的零样本泛化。然而，学习高性能通用策略需要采用Transformer（TF）等复杂架构，相比简单多层感知器（MLP），其内存和计算成本更高。为在推理时兼顾TF的优异性能与MLP的高效性，我们提出HyperDistill，包含：(1) 一种基于形态条件的超网络（HN），可生成面向特定机器人的MLP策略；(2) 一种对成功训练至关重要的策略蒸馏方法。在包含数百种多样化形态的基准测试UNIMAL上，HyperDistill在训练机器人和未见测试机器人上的性能与通用TF教师策略相当，但模型大小缩小6-14倍，不同环境中的计算成本降低67-160倍。分析表明，HyperDistill在推理时的效率优势源于知识解耦，即分离任务间与任务内知识的能力，该通用原则也可应用于提升其他领域的推理效率。