Real-time proprioception is a challenging problem for soft robots, which have virtually infinite degrees of freedom in body deformation. When multiple actuators are used, it becomes more difficult as deformation can also occur on actuators caused by interaction between each other. To tackle this problem, we present a method in this article to sense and reconstruct 3-D deformation on pneumatic soft robots by first integrating multiple low-cost sensors inside the chambers of pneumatic actuators and then using machine learning to convert the captured signals into shape parameters of soft robots. An exterior motion capture system is employed to generate the datasets for both training and testing. With the help of good shape parameterization, the 3-D shape of a soft robot can be accurately reconstructed from signals obtained from multiple sensors. We demonstrate the effectiveness of this approach on two soft robot designs—a robotic joint and a deformable membrane. After parameterizing the deformation of these soft robots into compact shape parameters, we can effectively train the neural networks to reconstruct the 3-D deformation from the sensor signals. The sensing and shape prediction pipeline can run at 50 Hz in real time on a consumer-level device.

中文翻译：

---

气动软体机器人 3D 变形的感知与重建

实时本体感觉对于软机器人来说是一个具有挑战性的问题，软机器人在身体变形方面几乎具有无限的自由度。当使用多个致动器时，由于彼此之间的相互作用也会在致动器上发生变形，因此变得更加困难。为了解决这个问题，我们在本文中提出了一种方法来感知和重建气动软机器人的 3D 变形，首先在气动执行器的腔室内集成多个低成本传感器，然后使用机器学习将捕获的信号转换为形状软机器人参数 采用外部运动捕捉系统来生成用于训练和测试的数据集。借助良好的形状参数化，可以根据从多个传感器获得的信号准确地重建软机器人的 3D 形状。我们证明了这种方法在两种软机器人设计上的有效性——机器人关节和可变形膜。在将这些软机器人的变形参数化为紧凑的形状参数后，我们可以有效地训练神经网络从传感器信号重建 3-D 变形。传感和形状预测管道可以在消费级设备上以 50 Hz 的实时频率运行。