Open Continuum Robotics -- One Actuation Module to Create them All

TL;DR

Open Continuum Robotics addresses the lack of a common torque-controlled continuum robot platform by introducing an open-source actuation module with backdrivable low-gear transmission and proprioceptive sensing via motor current. The module supports tendon displacement, backbone rotation, and tendon-induced motions, and is demonstrated through planar TDCR, spatial TDCR, and CTCR prototypes with high-rate control interfaces. Proprioceptive sensing of external forces is shown to be feasible, enabling future dynamics and interaction-aware control. By sharing CAD, hardware, and software, the work aims to enhance reproducibility, lower entry barriers, and catalyze research into advanced control and learning for continuum robots.

Abstract

Experiments on physical continuum robot are the gold standard for evaluations. Currently, as no commercial continuum robot platform is available, a large variety of early-stage prototypes exists. These prototypes are developed by individual research groups and are often used for a single publication. Thus, a significant amount of time is devoted to creating proprietary hardware and software hindering the development of a common platform, and shifting away scarce time and efforts from the main research challenges. We address this problem by proposing an open-source actuation module, which can be used to build different types of continuum robots. It consists of a high-torque brushless electric motor, a high resolution optical encoder, and a low-gear-ratio transmission. For this letter, we create three different types of continuum robots. In addition, we illustrate, for the first time, that continuum robots built with our actuation module can proprioceptively detect external forces. Consequently, our approach opens untapped and under-investigated research directions related to the dynamics and advanced control of continuum robots, where sensing the generalized flow and effort is mandatory. Besides that, we democratize continuum robots research by providing open-source software and hardware with our initiative called the Open Continuum Robotics Project, to increase the accessibility and reproducibility of advanced methods.

Figures (8)

• Figure 1: An actuation module applicable to generate tendon displacement/tension as well as for backbone rotation/torque for a wide range of continuum robots.
• Figure 2: Actuation module with coupling system. (left) Exploded view of a actuation module. (middle) Coupling system, where the second part can be changed on the fly. (right) Actuation module with different use cases for specific motion modes. To pull on a tendon, the tendon is terminated and spooled on the drum. For translational motion, a pinion gear is inserted. To rotate a backbone or tube, the tube or rod is attached to the coupling system, which has a through hole.
• Figure 3: Step responses of the actuation module with controllers tuned by the standard Ziegler-Nichols method. The controller gain $k_\text{P}$ is 1.0 for the P controller. The PD controller gains are $k_\text{P} = 1.6$ and $k_\text{D} = 0.02$, whereas the gains for the PID controller are $k_\text{P} = 0.9$, $k_\text{I} = 10.1$, and $k_\text{D} = 0.02$.
• Figure 4: Comparison of PD controllers with and without gravity compensation. The desired step is shown as a black dashed line. It is evident from the step responses that the higher the load, the higher the steady-state error of the naïve PD controller. Furthermore, a qualitative comparison shows a predictable pattern -- each steady-state error is (nearly) proportional to the load, which can be compensated by adding a constant current to the naïve PD controller leading to a PD-$g(\theta)$ controller. Step responses of different heights with different tendon tensions show that the PD-$g(\theta)$ controller successfully compensates the known load, while the steady-state errors of a naïve PD controller are relative to the load.
• Figure 5: Following a desired $\mathcal{C}^4$ smooth trajectory with and without load and gravity compensation. The uncompensated load appears as a constant offset in the error. The performance of the PD-$g(\theta)$ controller with load coincides whit the performance of the PD controller without load. Note that, in an unloaded case, both controller behave similarly.