An innovative locomotion principle for minirobots moving ... - CiteSeerX

Proceedingsof the 2002 IEEE lntemationalConference on Robotics 8 Automation Washington, DC May 2002

An Innovative Locomotion Principle for Minirobots Moving in the Gastrointestinal Tract L. Phee, A. Menciassi, S . Gorini, G. Pernorio, A. Arena, P. Dario Scuola Superiore Sant’Anna- MiTecWCFUM Lab Piazza Martiri della LibertA 33 - 56127 PISA, Italy e-mail : arianna@,sssur>.it conventional colonoscopes could be replaced by semiautonomous self-propelling robots which navigate the colon by pulling a very thin and flexible “service tail”: the tail should only be used for transportation of air, water and electrical energy from an external source while the robot propels itself into the colon. Several semiautonomous robots for colonoscopy were developed [4-61 and extensively tested in the authors’ laboratory. The earlier robots, which will be illustrated in the following section, were based on an inchworm locomotion principle. These devices demonstrated an intrinsic inefficiency in navigating collapsed bends in the colon. As such, a different locomotion principle based on “sliding clampers” was investigated and developed as illustrated in the Section 3. The fourth Section presents the implementation of the new locomotion principle in a miniaturised prototype. The discussion about the results of related in vitro and in vivo tests and the presentation of the future work are reported in Section 4 and Section 5.

Abstract This paper illustrates a mechanism specifically designed for locomotion in the wet, collapsible and tortuous human gastrointestinal (GI) tract (the colon in particular). Previous works peformed in the authors ’laboratorywere devoted to the fabrication of semi-autonomous inchworm locomotion devices for navigation in the colon; in this paper a firther analysis of limitations and problems of these devices has been performed. The main limitation consists of the poor eflciency of these devices to negotiate acute bends (due to what the authors termed as the “accordion effect’.) and, in general, to advance in the scarcely supported colon tissue. Thus a direrent approach to locomotion has been developed based on ‘‘sliding clampers ”. This locomotion principle has been implemented in a minirobot system and has demonstrated (both theoretically and experimentally) to be effective in reducing the “accordioneffect’’.

2. Inchworm locomotion in the gastrointestinal tract

1. Introduction Presently, many diagnoses of important pathologies are performed by exploiting minimally invasive techniques which allow medical doctors to introduce advanced endoscopes in the human body through natural orifices or small incisions. The diagnosis of colon cancer, which is one of the main causes of death in industrialized countries [l], is carried out using a colonoscope which has onboard a CCD camera, bundles of optical fiber for illumination, several working channels for air, water and miniature wire-actuated instruments for local treatment and biopsy [2]. The introduction and advancement of a conventional colonoscope into the patient’s colon is a difficult and tedious procedure for medical doctors and is generally uncomfortable for patients. Pain could even result when the colon is over-insufflated with air or over-stretched by the rather rigid colonoscope which has a diameter ranging between 13 mm and 19 mm.

Several locomotion mechanisms for semi-autonomous endoscopes, often inspired by minirobots for industrial inspection [7], have been developed by various researchers [8-131 in the world. The authors approached the problem of locomotion by exploiting an inchworm device which is particularly suited to unstructured or even hostile environments where wheels and tracks fail. An inchworm device is made up of basically two types of actuators: clamper and extensor. The clamper is used to adhere or clamp the device securely onto the locomotion environment while the extensor produces a positive displacement (known as the stroke, i.e. the difference in length of the extensor in its elongated and retracted phases). The simplest inchworm device consists of two clampers at its ends and one extensor in the middle. Figure 1 shows the gait sequence of forward propulsion.

The recent development of CMOS cameras [3], which do not require high illumination intensity, would allow the reduction in size and weight of the “tail” of the colonoscope. Flexibility of the scope would even be enhanced with a smaller “tail”. Given this situation,

Theoretically, the inchworm device should advance a distance equal to its stroke length after each cycle of the locomotive sequence. However, the authors’ past experimental results [14] have shown that losses can result due to the slippery, elastic and loosely constrained

0-7803-7272-7/021$17.00 0 2002 IEEE

1125

position in the GI tract with little or no forward advancement. The authors have termed this the accordion effect which is illustrated in Figure 3. This same phenomenon can also happen in the elongation phase but in this case, the folds would be formed distal to the distal clamper rather than in between the two clampers.

characteristics of the GI tract. In fact, it is attached to other surrounding organs by means of elastic mesenteries that allow the gut to slide freely in between the other organs. The inchworm configuration of Figure 1 would perform reasonably well in a straight path. But in an acute and unconstrained bend, the GI tract could easily ‘crumple’ up and be thrown into folds in between the two clampers during the retraction phase.

Re”

Figure 1. Schematic diagram illustrating the sequence of the inchworm locomotion principle. The shaded area on the distal and proximal clamping actuators indicates the active clamping states. Figure 3. Illustration of the accordion effect.

With reference. to Figure 2, this means that points A and A’ (on the GI tract) would remain more or less in the same position with respect to the proximal clamper before and after the retraction phase. Although the proximal clamper has advanced a distance AIt, the real advancement (with respect to the GI tract) is only All. A loss of stroke length A& results.

To better understand losses during locomotion, the authors proposed to break ,the inchworm process into 3 distinct features: elongation, retraction and clamping. The individual efficiency of each of these mechanisms contributes to the overall efficiency of the locomotion system, which can be represented by: where qe, qr and qc are the efficiencies of the elongation, retraction and clamping mechanisms respectively. These efficiencies can easily be lcalculated by comparing the theoretical performance of the mechanism with its actual performance during in vitro or in vivo experiments. For example, if the extensor is designed to extend 100 mm but due to the accordion effect it is observed to only advance 50 mm in an in vitro experiment, qe would be calculated as 0.5. Since q is directly proportional to each component, it is important to maintain high individual efficiencies for effective locomotion. The first prototypes developed in the authors’ laboratory were totally pneumatically actuated the extensor consisted of a bellow and the clampers were two hollow cylindrical structures with numerous holes on its surface. A supplied vacuum causes the intestine tissue to be ‘sucked’ around the clamper thus generating traction forces which adhere the clamper onto the GI tract [ 141.

Figure 2. Losses in stroke length during retraction.

The folds that result would be straightened during the elongation phase. Repeating the gait sequence would only result in the device elongating and retracting at the same

1126

The derived traction forces are governed by the equation:

and constrained at both its ends. Before inserting the prototype in the colon, all parameters were measured h d optimized as follows:

F=jlN where F is the traction force, p is the coefficient of friction and N is the normal force. N increases as the suction pressure increases. However, due to the low coefficient of friction p, the derived traction force is limited. In vitro experiments with explanted pig’s colon have shown that traction forces of up to 6 N were achieved with a vacuum pressure of -0.6 bar’ using a clamper with a diameter of 22 mm. It was also observed that undesirable ‘red spots’ appeared with suction pressures lower than -0.4 bar. These ‘red spots’ would become lesions when suction time is prolonged. For this reason, the applied vacuum pressure should not be lower than -0.4 bar. This limitation further decreases the maximum traction force that can be generated. As such, the authors concluded that clampers based just on suction were not appropriated for this application. In order to improve the grasping efficiency of the robot, a different clamping mechanism was conceived and it is being exploited in the latest version of the minirobots. Instead of using the vacuum to obtain traction forces, it could be used to cause the tissue to fall into the ‘jaws’ of a mechanical clamp (e.g. grippers, pincers, forceps) [15]. Having a prominent hold on the tissue, the ‘jaws’ of the clamp can easily close in for a positive grasp. A typical inchworm prototype with a central bellow for elongation and retraction and “suction+mechanical” clampers is shown in Figure 4.

Stroke length: 8 cm Time of 1 cycle of gait sequence: 32 s Theoretical speed 0.25 c d s The device was inserted in the colon and its performance was recorded both for the entire path and for each colonic tract (bend or straight path). The clamper was observed to grip the GI tract tissue very securely (no slippage). This variation in speed is largely due to the accordion eflect, which is more prominent in curved paths. Although the specimen GI tract was securely fixed at both ends, folds were clearly visible during the elongation and retraction phases of the gait sequence when the device passes very acute bends. However, the device managed to propel itself through the entire path with an efficiency of about 70%. This was calculated by comparing the average recorded speed with the theoretical speed. An in vivo experiment was carried out on a 35 kg male pig under general anaesthesia with the assistance and collaboration of a specially trained medical team in accordance to all the ethical considerations and the regulatory issues related to animal experiments. Prior to the experiment, the pig’s bowels were properly prepared for colonoscopy and was inspected by using a conventional colonoscope. 2 bends in the colonic tract were revealed during the inspection. The first, a gentle bend, was situated about 30 cm from the anus while the second, an acute kink, was situated about 50 cm from the anus. After the withdrawal of the colonoscope, the prototype was introduced manually about 10 cm into the pig’s anus. Upon activation of the gait sequence, the inchworm device propelled itself a distance of 40 cm into the colon with an estimated speed of 0.19 c d s . After which, its speed decreases and the device was observed to remain stationary 55 cm from the anus. The gait sequence was stopped and the device was retrieved by manually pulling its ‘tail’. A second inspection with the conventional colonoscope revealed reddish clamping marks on the colonic walls. The last clamp mark was situated a few centimeters after the second, more acute bend. This showed that the distal head of the device conformed and surpassed the second bend but remained in this position there after. The locomotion efficiency of the device was calculated to be 72% along the straight portion of the colon; then it decreased to zero during the navigation of the second curve. As such, the main reason for its failure to surpass the second (less constrained) bend is due to the accordion eflect. In the in vitro experiment, the GI tract was well constrained and the device still managed to surpass the acute bends despite of existence of the accordion eflect. However, in the in vivo experiment, the

Figure 4. Inchworm prototype with “suction + mechanical ’’ clampers and a central bellow extensor (24 mm in diameter and has lengths of 115 mm and 195 mm when retracted and elongated respectively). 2.1 In vitro and in vivo experiments

In vitro experiments with this inchworm prototype was carried out with a “home-made” simulator made of polystyrene. This artificial path was pattemed according to the indications of medical doctors in order to reproduce a realistic 3-dimensional structure of the human colon. The overall size of the simulator was 60 cm x 90 cm x 20 cm and it included 60 cm of straight path and 105 cm of realistic “colon-like” curved paths. A freshly explanted pig’s colon of length 150 cm was placed in the test bench The more negative the pressure, the higher the suction power.

1127

pairs of clampers has been conceptually described. However, it is obvious that more pairs of clampers can be employed to achieve a more secure grip onto the GI tract.

GI tract (in its natural state) was not as well constrained. The accordion effect, in this case, was much more severe. From these experiments, it can be concluded that the accordion effect is intrinsically related to the position of clampers (at the ends of the robot) and extensor (in the middle). High locomotion efficiency cannot be derived from this disposition in a terrain which is not rigidly supported.

4. Minirobot with sliding clampers This prototype realizes the conceptual design described in Section 3. Figure 6 shows a 3-D model as well as a photo of the prototype. Having a inaximum O.D. and length of 16 mm and 35 mm respectively, it takes the hydrodynamic shape of a b,arrel to ensure easy insertion as well as withdrawal from the anus.

3. Locomotion principle based on “sliding clampers” In order to reduce the accordion eflect, a device was configured with two pairs of clampers working antagonistically such that one of which always grasp onto the intestinal tissue that is distal with respect to the other clamping mechanism. Figure 5 shows the working principle of this “sliding clampers” device advancing forward.

Two adjacent pairs of claimpers work antagonistically; when one pair is at the distal position, the other would be at the proximal position. The clampers protrude from the capsule only at the ends of the device. This feature reduces the friction generated from contact with the GI tissue during the sliding movements. The movements of the clampers, both for clamping and translation, are actuated by push-pull cables. Powered by motors, these clampers move in a sequence regulated by a computer software to bring about locomotion of the device.

(a) 00 Figure 6. (a) 3 0 model and (b) photo of sliding clampers prototype.

Figure 7 shows the arrangement of the push-pull cables within the prototype. A single cable, strung four times across the length of the prototype, is responsible for the translation motion of the clampers. The clamping action of each of the four clampers is actuated by four separate cables thus allowing individual clamping actuation. An air tube fixed to the internal cavity of the prototype delivers the required vacuum and positive air pressures to facilitate clamping. P-

Figure 5. Schematic diagram illustrating the sequence of the “sliding clampers ” locomotion principle. The shaded area indicates the active clamping states.

Clamp Pair A, shown to be initially distal with respect to Clamp Pair B, is activated to grasp the GI tissue. Both pairs of clamps then slide simultaneously such that Clamp Pair B is now distal with respect to Clamp Pair A. Clamping Pair B is then activated to grasp the GI tissue while Clamping Pair A releases its grip. A repetitive sequence of these phases can ensure that the clamping mechanism always grasp the tissue distal to its predecessor, thus reducing the accordion effect as the device propels itself forward. A device with only two

\

\

. \

Push-pull cable

Figure 7. Actuation of sliding clampers.

1128

Clamper

c

A user-friendly Human Machine Interface (HMI) and a data acquisition system have been designed so that the operator can activate the prototype and change working parameters (e.g. air pressures, clamping time, clamping force) on-line. With a stroke length nearly equal to its body length, this prototype has the potential of high speeds as compared to an inchworm minirobot of similar dimensions. In fact, the ratio of stroke length over body length of this device (0.63) is greater than that of the inchworm prototype (0.41).

compensates for this lost of efficiency. However, in the case of the sliding clampers prototype, which has a much shorter stroke length (2.2 cm), this lost in stroke length has a more drastic effect on its locomotion efficiency. PathDescri tion S eed cm/s Straight, 3D path, hump height 0.053 3D curve of 80°, curvature radius 12 cm,hum hei t 5 cm 2D curve of 180°, curvature 0.03 radius 10 cm I 2D curve of 8 4 O . CurvatureI 0.046 radius 12 cm I Mix path 0.039 Table 1.Actuation of sliding clampers.

4.1 In vitro and in vivo experiments

The same in vitro experimental setup used to evaluate the inchworm prototype (Section 2.1) has been used for the testing of the sliding clampers prototype. The optimized working parameters of the device are as follows:

The device was observed to propel itself with quite a constant speed along the entire simulated path. Folds are still visible but most of which are formed proximal to the device as depicted in Figure 8.

,,

25%

I I I

I

19% 29%

25%

As with the in vitro test, a similar in vivo experimental setup (as described in Section 2.1) was prepared to evaluate the performance of the sliding clampers prototype. The test specimen was a 30 kg female pig under complete anaesthesia. The ten cables and one air tube which exit the proximal end of the prototype was bundled as one “tail” to ease the process of insertion. The prototype moved at a constant speed for about 40 cm (measured from outside) before slowing down and finally coming to a halt. The experiment ended after about 15 minutes when no m h e r locomotion was visible. The pig’s abdomen was opened and the intestine was extracted with the intention to inspect for possible damages of the tissue. The robot’s position indicated that it overcame two intestinal bends and it stopped after a very acute bend located between the ascending and the transverse colon. The locomotion efficiency, although lower than that of the inchworm prototype, was constant throughout the entire experiment. Since the entire device halted after the bend, it can be concluded that the accordion efect, although it existed, did not cause locomotion to end. The main cause of stoppage was likely due to the ten steel push-pull cables that formed the ‘tail’ of the prototype. The high stiffness that resulted, as these are bundled together as one ‘tail’, caused too much friction against the GI wall. The propulsion force of the prototype was insufficient to overcome the excessive frictional forces generated from pulling the ‘tail’.

Stroke length: 2.2 cm Time of 1 cycle of gait sequence: 14 s Theoretical speed 0.16 c d s

Direction of locomotion

Efficienc 34%

I

Figure 8. Formation offolds proximal to the device. The accordion effect still existed and it caused a reduction in speed. However its speed was quite independent of the bends in the GI tract, thus demonstrating that the new locomotion mechanism can propel better in unconstrained and curved colon tissues than the inchworm system. The main results of the in vitro tests are reported in Table 1. Although the efficiency of the sliding clamper device is lower than that of the inchworm prototype, it has demonstrated that it is capable of reducing the accordion effect. This drop in locomotion efficiency can be explained as follows. Due to the high elasticity of the GI wall, the clamper, in its deactivated state, causes it to stretch as the former moves forward. The more the GI wall is stretched, the stiffer it becomes. There comes a point it will start to slip from the surface of the clamper. This is when the effective stroke begins. In the case of the inchworm prototype, its long stroke length (8 cm)

5. Conclusions The mechanical configuration of the device with sliding clampers is intrinsically more effective for locomotion in the collapsible and unconstrained intestine: the principle of “pushing back the intestine tissue”, according to our model, assures a uniform locomotion into the colon. On the other hand, comparing the in vitro results of both prototypes, it is clear that the inchworm prototype has a higher speed and locomotion efficiency with respect to

1129

[5] P. Dario, M.C. Carrozza, A. Pietrabissa, B. Magnani and L. Lencioni, “Endoscopic robot,” United States Patent, patent number 5,906,591, May 1999. [6] A. Menciassi, A. Arena, L. Phee, D. Accoto, C. Stefanini, G. Pemorio, S. Giorini, M. Boccadoro, M.C. Carrozza, P. Dario, “Locomotion Issues and Mechanisms for Microrobots in the Gastrointestinal Tract,” Proceedings of the 32nd ISR (Intemational Symposium on Robotics), Seoul, Korea, 19-21 April 2001, pp 428432. [7] A. Sadamoto, H. Sudo, €I. Yamada, T. Togazaki, M. Kimura, N. Kawahra, K. T S L U UT. ~ ~Shibata, , H. Izu and T. Sakakibara, “Wireless Micromachine for In-pipe Visual Inspection and the Possibility of Biomedical Applications,” Proceedings of the 32nd ISR (International Symposium on Robotics), Seoul, Korea, 19-21 April 2001, pp. 433-438 [8] T. Fukuda, H. Hosokai and M. Uemura, “Rubber gas actuator driven by hydrogen storage alloy for in-pipe inspection mobile robot with flexible structure,” in Proc. 1989 IEEE Int. Con$ Robot. Automat., Scottsdale, AZ, May 1989, pp. 1847-1852. [9] K. Ikuta, M. Tsukamoto and S. Hirose, “Shape memory alloy servo actuator system with electric resistence feedback and application for active endoscope,” in Proc. 1988 IEEE Int. Con$ Robot. Automat., Philadelphia, Pennsylvania, April 1988, pp. 427-430. [lo] M. R. Treat and W. S. Trimmer, “Self-propelled endoscope using pressure driven linear actuators,” United States Patent, patent number 5,595,565, Jan. 1997. [ l l ] W. S. Ng, S. J. Phee and F. Seow-Choen, “Robotic endoscope and an autonomous pipe robot for performing endoscopic procedures,” United States Patent, Patent Number 6,162,171, Dec. 2000. [12] W. S. Grundfest, J W Burdick and A B Slatkin, “Robotic endoscopy,” United States Patent, patent number 5,662,587, Sep. 1997. [13] S. J. Phee, W. S. Ng, I. IM.Chen and F. Seow-Choen, “Development of new locomotive concepts to be used in automation of colonoscopy,” in 9th Int. Con$ for BioMedical Engineering, Singapore, Dec. 1997, pp. 8792. [14] M. C. Carrozza, L. Lencioni, B. Magnani, S. D’Attanasio and P. Dario, “The development of a microrobot system for colonoscopyYy’ in Proc. First Joint Con$ of Computer Vision, Virtual Reality and Robot. in Medicine and Medical Robot. and Computer Assisted Surgery, vol. 1205, pp. 779-’788, Springer-Verlag, BerlinHeidelberg, 1997. [15] S.J. Phee, A. Arena, A. Menciassi, P. Dario, “Endoscopic device for locomotion through the gastrointestinal tract”, International patent application PCT/KR01/00304 (pending).

the sliding clampers prototype. The higher speed of the former prototype is due to its larger size which enables it to have a much longer stroke length. In the in vitro situation, the GI tract was securely constrained, reducing the accordion effect. In the in vivo experiments however, the colon (in its natural state) was not as rigidly constrained. The accordion effect prevented the inchworm prototype from surpassing the second more acute bend, reducing its locomotion efficiency to zero in the process. The sliding clampers prototype, despite its lower speed, managed to surpass the anatomically identical acute bend maintaining its low but constant locomotion efficiency, as indicated by our theoretical model. In doing so, the sliding clampers prototype has proven its capability in reducing the accordion effect. The authors’ tbture works include the development of a minirobot with the good attributes of both locomotion concepts: high speed, effectiveness in bends and unconstrained intestine, low rigidity and small dimension. A tilting head to help locomotion in bends, a microcamera with light source, on-board actuators and other micro-endoscopic instruments and sensors would also be integrated into the minirobot in the near tbture in order to fabricate a hctional device.

Acknowledgments The authors wish to acknowledge that this paper is a result of the research accomplished with the financial support of the Intelligent Microsystem Center, Seoul, Korea, which is carrying out one of the 21st century‘s New Frontier R&D Projects sponsored by the Korea Ministry of Science & Technology (www.microsystem.re.kr). The authors wish to thank M.G. Trivella, MD, Prof. P. Spinelli, MD, Prof. A. Pietrabissa, MD and Prof. S.Y. Song MD for their medical support and consultancy.

References [l] B. A. Miller, L. N. Kolonel, L. Bemstein, Jr. J. L. Young, G. M. Swanson, D.West, C. R. Key, J. M. Liff, C. S . Glover and G. A. Alexander, “RaciaUEthnic patterns of cancer in the United States 1988-1992,’’ National Cancer Institute, NIH Pub. No. 96-4104: Bethesda, MD, 1996. [2] P.B. Cotton, C.B. Wiliams, “Practical Gastrointestinal Endoscopy”, Oxford, Blackwell Scientific Publication, 1990. [3] G Iddan, G Meron, A Glukhovsky and P. Swain, “Wireless Capsule Endoscopy”, Nature, vol. 405, no. 6785, p. 417,2000. [4] P. Dario, M. C. Carrozza and A. Pietrabissa, “Development and in vitro tests of a miniature robotic system for computer assisted colonoscopy,’y Journal of Computer Aided Surgery, vol. 4, pp. 4-14, 1999.

1130