Image-guided control of a continuum robot

Janssen, S.L. (2012) Image-guided control of a continuum robot.

Abstract:This study presents the development of a continuum robot which could serve as part of a surgical robot. Continuum robots, because of their dexterous nature and continuous shape are suited for minimally invasive surgery. Their inherent flexibility also reduces the risk of damage. Continuum robots have these properties due to the absence of discrete joints. Previous research in the domain of continuum robots has focused on developing kinematic and dynamic models. These models attempt to take in account as much of the inherent imperfections such as friction and fabrication errors as possible. This study instead focussed on showing that a simple kinematic model based on the ideal behaviour of the robot is sufficient when combined with a closed-loop controller with 3D position feedback. A continuum robot and its actuator have been developed. The continuum robot was tendon actuated. It consisted of a flexible nitinol backbone with four disc shaped tendon guides fixed equidistantly to each other to it. Two designs of robot were created: A 160 mm length robot with 5 mm radius tendon guides and a 160 mm one with 10 mm radius tendon guides. The actuator was fitted with four motors to pull the tendons and had a translation stage for an insertion movement. A simple kinematic model based on the constant curvature principle was developed. This principle assumes that the backbone bends in a circular arc. A PID-controller was implemented. This PID-controller was modified to improve performance with non-continuous input signals. A stereoscopic camera setup provided the feedback for this controller. An image processing algorithm took the images provided by the cameras and triangulated the 3D position of the last tendon guide. The controller, kinematic model and image processing were implemented as a multi-threaded C++ win32 application. The complete system worked at 15 Hz. Experiments were performed with the 20 mm radius design to compare the performance of the open-loop system with the closed-loop system. Every experiment was performed 5 times. While the open-loop system had a 60 mm step response steady state error of approximately 15 mm, the closed-loop system had a steady state error of about 2 mm. 3D shapes were used as input as well. The most complex of these shapes was a series of squares that were sequentially shifted in space, forming a beam shape. The open-loop system had an average tracking error of 20 mm while tracking this shape. The closed loop system had an average tracking error of 6 mm. A 3D haptic device was also used as an input for the system. The system was capable of tracking the input of a human following a cross shape pattern with an average tracking error of approximately 4 mm. The 5 mm radius robot was also tried, but due to excessive friction the robot could not be controlled in a stable manner. This study showed that a simple kinematic model combined with a 3D position feedback controller is an alternative for a more complex kinematic model. Future research involves embedding fibre Bragg grating strain sensors in the backbone of the robot. This would allow the reconstruction of the robot shape in 3D while in vivo.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:53 electrotechnology
Programme:Electrical Engineering MSc (60353)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page