Search papers, labs, and topics across Lattice.
The paper introduces WestWorld, a trajectory world model designed to scale across diverse robotic systems by incorporating system-specific and structural knowledge. To achieve scalability, WestWorld employs a system-aware Mixture-of-Experts (Sys-MoE) that dynamically routes specialized experts based on a learned system embedding. By further incorporating a structural embedding that aligns trajectory representations with morphological information, WestWorld achieves improved zero-shot generalization and downstream model-based control, validated through experiments including real-world deployment on a Unitree Go1.
Forget training individual models for each robot: WestWorld learns a single, scalable trajectory world model that generalizes across 89 diverse robotic systems by encoding both system-specific dynamics and physical structure.
Trajectory world models play a crucial role in robotic dynamics learning, planning, and control. While recent works have explored trajectory world models for diverse robotic systems, they struggle to scale to a large number of distinct system dynamics and overlook domain knowledge of physical structures. To address these limitations, we introduce WestWorld, a knoWledge-Encoded Scalable Trajectory World model for diverse robotic systems. To tackle the scalability challenge, we propose a novel system-aware Mixture-of-Experts (Sys-MoE) that dynamically combines and routes specialized experts for different robotic systems via a learnable system embedding. To further enhance zero-shot generalization, we incorporate domain knowledge of robot physical structures by introducing a structural embedding that aligns trajectory representations with morphological information. After pretraining on 89 complex environments spanning diverse morphologies across both simulation and real-world settings, WestWorld achieves significant improvements over competitive baselines in zero- and few-shot trajectory prediction. Additionally, it shows strong scalability across a wide range of robotic environments and significantly improves performance on downstream model-based control for different robots. Finally, we deploy our model on a real-world Unitree Go1, where it demonstrates stable locomotion performance (see our demo on the website: https://westworldrobot.github.io/). The code will be available upon publication.
