Current medical robotics systems use non-intelligent surgical manipulators that place the entire burden on surgeons for safeguarding against damage to the anatomy. The emergence of new surgical paradigms, such as Natural Orifice Endoscopic Trans-luminal Surgery (NOTES), requires surgical robots that are capable of supporting safe interaction with the anatomy while accessing deep surgical sites through often long natural access pathways. This requires new types of robots capable of safeguarding against damage to the anatomy by acting as intelligent intervention and information gathering tools for assisting surgeons during increasingly complex procedures.
The objective of this research is to provide the theoretical foundation for modeling and control of flexible robots for intelligent and safe interaction with the anatomy. Intelligence refers to the ability of these robots to gauge their force interaction with the anatomy, gather information about the anatomy, and act based on this information. Screw theory and stochastic estimation methods are used for modeling the ability of these robots to estimate their wrench interaction with the anatomy by using intrinsic and extrinsic sources of information. These performance measures are used in hybrid force control algorithms that allow characterizing shape, stiffness, and anatomical constraints governing safe maneuvering of suspended organs.
The outcomes of this research will allow the development of radically new technologies for newly emerging surgical paradigms (e.g. NOTES). This research will also advance the field robotics by addressing control and resolution of multi-point contact problems along flexible robots for compliant insertion control and bracing against soft environments.
Current surgical systems cannot traverse complex anatomical passages, have no sensory feedback and no intelligence. The emergence of new surgical paradigms, such as natural orifice surgery, demands surgical robots capable of supporting safe interaction with the anatomy while accessing deep surgical sites through sinuous natural access pathways. The goal of this research was to provide a theoretical foundation for modeling and control of continuum (snake-like) robots for intelligent and safe interaction with the anatomy. Continuum robots are flexible robots capable of achieving sinuous shapes by actuating their flexible structural members. We developed a modeling framework for the kinematics, statics and force sensing of continuum robots. We also developed a framework allowing them to be used for manipulation and force/moment (i.e. wrench) sensing. By measuring the actuation forces, we created an intrinsic force sensing framework. Using this framework, we demonstrated palpation of flexible anatomy and generated estimated stiffness maps of the tissue. Figure 1 shows a stiffness map generated by the palpation process based on intrinsic force sensing. While this approach allowed sensing of interaction forces at the robot tip, it did not resolve the problem of detecting multiple contacts along the entire length of the robot. Solving this problem is important because it allows these robots to be safely inserted into an unknown environment or anatomic passageway where multiple contacts along the robot length are unavoidable. We therefore investigated two approaches for contact detection. The first approach uses information from load cells measuring actuation forces. The second approach uses the change in the movement of these robots as a result of constraining contact with the environment. Our investigation found the limitations of contact detectability using each one of the two methods mentioned above. A by-product of these two approaches was an extended general kinematic model that takes into account constrained segments or sub-segments. Using this model in a least-squares estimation framework, we were able to estimate the location of the contact along the length of the snake robot within an error of one to two spacer disk spacing (typically around 2-3 mm). While the demonstration of this algorithm used multi-backbone robots such as the one in figure 1 our modeling method extends to any type of continuum robots. To enable safe insertion of these robots along sinuous and unknown passageways there is a need for active compliant motion control methods that allow rapid and safe deployment of these robots in unknown environments. We therefore investigated and developed a new approach for modeling the stiffness of these robots and using this approximate stiffness estimate, we developed an approach for compliant motion control of these robot. This approach uses estimates of the robot compliance and the loads on the robot backbones to command the robot movement so as to minimize the force interaction of the robot with its environment. This approach was then validated in both rigid environments and in scenarios inspired by natural orifice surgery. Figure 2 shows one such experiment where the robot is changing its shape so as to actively allow insertion into an unknown acrylic tube. We also developed a modeling and sensing framework for estimating the characteristics of the environment using two approaches. In the first approach the robot induces an excitation of the tissue and uses an adaptation of a previously known method to estimate the perceived environment stiffness, impedance, mass and damping. In the second approach the robot attempts exploratory manipulation movement and uses its force sensing capability to estimate the perceived environmental constraint. The first method showed that it can enable semi-autonomous exploration of deep clefts in flexible anatomy. The second method showed that our modeling approach can be augmented with artificial intelligence classification methods to estimate the type of environmental constraint. We also developed a new method for hybrid force and position control of continuum robots using intrinsic force sensing. We demonstrated that this hybrid control method can be used to allow robots to assist in safely tele-manipulating the robot tip while regulating force on the environment. These fundamental breakthroughs in modeling and control of continuum robots have allowed the creation of new technologies for surgery. As this technology matured, we have synergistically investigated the utility of these robots and algorithms for investigating applications of these robots to single port access surgery, transurethral surgery and throat surgery. Figure 3 shows examples of these robots. This research has produced fundamental modeling results allowing the creation of new enabling technology for creating robots capable of being rapidly and safely deployed in unknown environments. Sample applications of these robots include search and rescue, inspection of pipes and surgical assistance for NOTES. This research has resulted in new technologies that have been licensed to several companies that are designing future-technologies for surgical assistance.
