面向敏捷腿足运动的状体估计Transformer
State Estimation Transformers for Agile Legged Locomotion
AI总结：
本文提出State Estimation Transformers（SET），将状态估计建模为条件序列建模，利用因果掩码Transformer预测四足机器人身体高度和速度等特权状态，在Cyberdog2上验证了其在跳跃、后空翻和侧滑任务中的估计精度、可迁移性及真实环境成功率优势。
摘要：
我们提出一种状态估计方法，能够准确预测机器人的特权状态，以突破四足机器人在野外执行跳跃等高级技能时的极限。具体而言，我们提出了State Estimation Transformers（SET），这是一种将状态估计问题建模为条件序列建模的架构。SET利用因果掩码Transformer，输出在现实世界中难以直接获取的机器人状态，例如身体高度和速度。通过将自回归模型以机器人过去的状态为条件，我们的SET模型即使在高度动态的运动中也能准确预测这些特权观测。我们在低成本四足机器人Cyberdog2上对三个任务进行了评估——奔跑跳跃、奔跑后空翻和奔跑侧滑。结果表明，SET在仿真中的估计精度和可迁移性，以及在现实世界中跳跃成功率和触发恢复控制器的成功率方面均优于其他方法，表明这种基于Transformer的显式状态估计器在高度动态运动任务中具有优越性。
英文摘要：
We propose a state estimation method that can accurately predict the robot's privileged states to push the limits of quadruped robots in executing advanced skills such as jumping in the wild. In particular, we present the State Estimation Transformers (SET), an architecture that casts the state estimation problem as conditional sequence modeling. SET outputs the robot states that are hard to obtain directly in the real world, such as the body height and velocities, by leveraging a causally masked Transformer. By conditioning an autoregressive model on the robot's past states, our SET model can predict these privileged observations accurately even in highly dynamic locomotions. We evaluate our methods on three tasks -- running jumping, running backflipping, and running sideslipping -- on a low-cost quadruped robot, Cyberdog2. Results show that SET can outperform other methods in estimation accuracy and transferability in the simulation as well as success rates of jumping and triggering a recovery controller in the real world, suggesting the superiority of such a Transformer-based explicit state estimator in highly dynamic locomotion tasks.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
