Search

Haptic feedback is typically missing during telemanipulation of surgical robots in minimally invasive surgeries, i.e., surgeons cannot feel the interaction forces between the instruments and tissues. Instead, surgeons have to solely rely on visual feedback, which increases complexity of guiding the instruments and poses the safety threat of unperceivable contacts outside the field of view. We propose a novel series elastic actuation design for articulated robotic endoscopes to overcome these limitations and evaluate an according device with one joint. Similar to the actuation of human fingers, the joint is driven by antagonistic tendons. Springs are integrated in the transmission between the tendons and the motors outside of the endoscope shaft. We estimated the joint angle and thereby the endoscope shape, measured spring deflection, estimated tendon forces from that deflection, and implemented force control for the endoscope joint. Zero torque control and impedance control were evaluated under application of both a continuous force and an impact force to the endoscope tip. The springs reduced impact forces at the tip of the endoscope through their inherent compliance. At the same time, feeding back the estimated force resulted in a stable tendon force control and a tunable endoscope joint control: Zero torque control effectively reduced the external forces, while the endoscope joint showed the expected stiffness in impedance control. These results show that antagonistic series elastic actuation is a promising concept for endoscope joint actuation and that it can lead towards safer robot–tissue interactions in surgical robotics.

While telemanipulating a surgical robot, surgeons usually miss haptic feedback, i.e., they cannot feel interactions between the surgical instruments and the tissue. The lack of haptic feedback makes the surgeons solely dependent on visual information, which might render the handling of surgical robots unintuitive and thereby poses a potential safety threat. To overcome these limitations, we propose to use Series Elastic Actuation (SEA) for tendon-driven endoscopes. In this regard, we developed an endoscope prototype with SEA to demonstrate its potential for safer robot-tissue interactions: A spring was integrated into the tendon transmission between the motor and the movable endoscope tip. The deflection of the spring was measured and used to estimate the tendon force. The estimated tendon force was then fed back to a force controller. Our tendon force estimation had lower noise levels than that of a commercial force/torque sensor, but showed a systematic error, which we assume to originate from friction. In response to an input step, the tendon force settled to the desired value within 50 ms and the steady-state error was below 0.05 N. These results showed that SEA is a promising concept for force measurements and robust force control in tendon-driven endoscopes.

This work was done in the framework of the MIRACLE project devoted to the development of a flexible, articulated robotic endoscope to perform laser osteotomies (bone cutting). In order to control this robot with the required accuracy of 0.2 mm at the tip position, a tracking system needs to feed back the pose of the tip and the shape of the endoscope. One of the possible designs of such a robotic endoscope is based on an articulated device that can be tracked by measuring the bending angle at each joint. Recently, we presented (Iafolla et al., 2018) a novel angular sensor called ASTRAS (Angular Sensor for TRAcking Systems), which measures a one degree of freedom rotation by processing a shadow image cast through a shadow mask onto an image sensor. One ASTRAS will be placed at each joint of the articulated robotic endoscope in order to measure their bending angles. The planned diameter of the MIRACLE endoscope is 10 mm and it will contain several components such as the optical fiber for the laser, irrigation and sucking systems, and robot control mechanisms. However, the latest version of ASTRAS covered a volume of 40x40x55 mm3. For this reason, the most demanding requirement for the design of the final ASTRAS sensor is its miniaturization to fit the sensor inside the joint of the endoscope. Consequently, in this contribution, we investigated the possibility of miniaturizing ASTRAS and assess its limitations due to some fundamental aspects. In particular, we built a prototype based on a NanEye camera by AMS (1x1x0.5 mm3) and a micro-milled shadow mask with 0.08 mm wide slits. The tests of this prototype showed encouraging results and will be discussed. In particular, at such small dimensions, there are no unexpected optical effects (e.g. optical diffraction due to the mask) worsening the measurement. Indeed, the precision of 5 arcsec and the linearity error of 23 arcsec RMS are comparable to those of the previous bigger prototype. Further considerations and concepts related to the miniaturization of ASTRAS are also discussed in this paper.