Joint Goal and Strategy Inference across Heterogeneous Demonstrators via Reward Network Distillation