Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Medical robotics simulation framework for application-specific optimal kinematics

Medical robotics simulation framework for application-specific optimal kinematics Current Directions in Biomedical Engineering 2019;5(1):145-148 Sven Böttger*, Tolga-Can Çallar, Achim Schweikard and Elmar Rueckert Medical robotics simulation framework for application-specific optimal kinematics Abstract: Most kinematic structures in robot architectures Keywords: medical robotics, robot kinematics, optimization, for medical tasks are not optimal. Further, the workspace anthropometric body shape data and payloads are often oversized which results in high https://doi.org/10.1515/cdbme-2019-0037 product prices that are not suitable for a clinical technology transfer. To investigate optimal kinematic structures and configurations, we have developed an adaptive simulation framework with an associated workflow for requirement 1 Introduction analyses, modelling and simulation of specific robot kinematics. The framework is used to build simple and cost For medical robotic applications, standard industrial robots are effective medical robot designs and was evaluated in a tool often used because of their good commercial availability, manipulation task where medical instruments had to be product quality, and accuracy, although they do not optimally positioned precisely and oriented on the patient's body. The meet the kinematic requirements for the application. These model quality is measured based on the maximum systems are designed for universal use for a variety of tasks in workspace coverage according to a configurable scoring various industries and are therefore usually oversized in metric. The metric generalizes among different human workspace and payload, whilst also being expensive and body shapes that are based on anthropometric data from requiring special security measures. UMTRI Human Shape. This dexterity measure is used to Some related studies that investigate optimal kinematic analyze different kinematic structures in simulations using structures in medical robotics exist. Yoshikawa [1] discussed the open source simulation tool V-REP. Therefor we the manipulating ability of robotic mechanisms in positioning developed simulation and visualization procedures for and orienting end-effectors and proposes a measure of medical tasks based on a patchwork of size-variant manipulability. Some performance measures were reviewed anatomical target regions that can be configured and by Patel [2]. Paden [3] formulated an optimality theorem for selectively activated in a motion planning controller. In our six revolute joints kinematics and Nelson [4] proposed a evaluations we compared the dexterity scores of a Monte Carlo simulation algorithm for optimizing a redundant commercial lightweight robot arm with 7 joints to serial spherical linkage. Pamanes [5], Zeghloul [6] and optimized kinematic structures with 6, 7 and 8 joints. Com- Vidaković [7] presented methods to find the optimal pared to the commercial hardware, we achieved placement of robots. Xiang [8] proposes a three-dimensional improvements of 59% when using an optimized 6- space path prediction simulation method and a design process dimensional robot arm, 64% with the 7-dimensional arm method for robotic medical tool-guidance manipulators was and 96% with an 8-dimensional robot arm. Our results proposed by Nouaille [13]. show that simpler robot designs can outperform the In this work, we perform the analysis of the kinetic typically used commercial robot arms in medical requirements, in particular of workspace and dexterity, applications where the maximum workspace coverage is covering all the above issues. Furthermore, we determine essential. Our framework provides the basis for a fully design and a configuration of an optimal robot, especially for automatic optimization tool of the robot parameters that can applications in the medical field for the manipulation of be applied to a large variety of problems. instruments on the human body, e.g. robot-assisted ultrasound or needle puncture. ______ *Corresponding author: Sven Böttger: e-mail: 2 Methods boettger@rob.uni-luebeck.de Tolga-Can Çallar, Achim Schweikard, Elmar Rückert: Institute for Robotics and Cognitive To analyse, model and simulate the application-specific Systems, Universität zu Lübeck, Ratzeburger Allee 160, Lübeck, optimal robot kinematics we developed and evaluated a Germany Open Access. © 2019 Sven Böttger et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 License. S. Böttger et al., Medical robotics simulation framework for application -specific optimal kinematics — 146 workflow and a software framework. The application-specific range and axial offset of the joints and the arm link lengths) target workspaces were modelled and compared with the robot and the design of the tool geometry of the application-specific workspaces, and the score values were determined according instrument given by the tool-tip coordinates. The general to a dexterity metric, and hence, a manual optimization was functional requirements for the kinematic design were carried out. evaluated using the design process by Siciliano [9]. The design goal is a simple, lightweight, dynamically stable and cost- effective robotic arm. The structure with as few joints as 2.1 Computing targets using anatomical possible represents the most important optimization criterion that meets all goal criteria. body models The first section of the workflow was the requirement analysis. We utilized a statistically representative anatomical 2.3 Dexterity simulation body model based on anthropometric data containing some In the simulation, as the third workflow section, the two model variations regarding size and shape to cover the spectrum of components were integrated into a common simulation human anatomy (section Software and simulation framework). environment and virtually interconnected, whereby the kinetic The target areas for the intended medical applications were performance was determined. A test algorithm has been marked by selecting vertices from the surface of the body implemented to approximate the functional requirements of model. Likewise, different regions can be combined with each the intended medical application and tests the kinematics using other depending on the application scenario. a dexterity metric defined as follows: It tests whether the In a second workflow step, the models for the target instrument could be virtually positioned at all target positions workspace and the robot workspace were modelled in parallel. and oriented in some different rotations (next section) in The target workspace was created by generating multiple body parameterized Tait–Bryan (roll-pitch-yaw) angles. Target model variations by adjusting the physiognomic parameters positions for testing are all selected vertex points contained in for gender, patient size, body mass index, body length, and age all layers of the target workspace model. Those were assigned to produce a minimal and maximal shell model of the body with direction vectors normal to the model surface. The base surface. Based on the DINED Anthropometric Database [1], th of the robot model was positioned relative to the target model this range was selected from the 5 percentile (corresponding in the simulation space, whereby the position has significant to the body height of 1.54 m with a body mass index [BMI] of th impact on the result and is also subject to optimization. The 20) to 95 percentile (1.91 m, BMI = 34) of the anatomical dexterity estimation was implemented using an algorithm for bandwidth (see Fig. 1a). inverse kinematics calculation and path planning (section The generated model data follows an identical general Software and simulation framework), which attempts to find vertex structure, only differing in spatial deformation all target configurations on collision-free paths from the according to the input parameters. Subsequently, the mapping starting position to the target position. Self-collisions and of the application-specific target areas to the likewise (layer-selective) collisions with the phantom must be avoided. application-specifically dimensioned shell models is performed by assigning the vertex indices. In the following morphing between minimum and maximum shell any number of intermediate shells can be created, controlling the accuracy of the dexterity calculation. The superposition of all vertex points belonging to the target areas from all shells yields the spatial target workspace. Finally, the number of vertex points is reduced to an appropriate ratio (in the present experiment to 1533) in order to reduce the computational demands in the th th Figure 1: (a) Anthropometric shell model generation of 5 and 95 simulation. percentile of the population and an intermediate shell by morphing operations (1), marking of grouped colored target areas (2), cropping (3) and superposition in supine (4, frontal 2.2 Optimizing kinematic structures and lateral view) yields the 3D target workspace embracing solely the colored vertices. (b) Discretization scheme of 5 orientations of the medical instrument for dexterity test to apply on every vertex position in target space relative to their We varied all the relevant kinematic parameters that form the direction. workspace of an articulated robot arm (namely number, types, S. Böttger et al., Medical robotics simulation framework for applicatio-nspecific optimal kinematics — 147 Figure 2: (a) Virtual simulation scene including target workspace phantom and 6-Joint kinematic. (b) Design and dimensions of the self-designed kinematic structures.. For planning of the collision-free paths to the target configurations, a sample based tree planner search algorithm (section Software and simulation framework) operating in Figure 3: Numerical dexterity distribution of the four examined joint parameter space is used. It selects always the shortest robot kinematics including mean dexterity score and standard path out of the generated plans. deviation. In each experiment 1533 targets had to be reached. Netherlands), highlighting the vertex groups belonging to the target areas within the polygon models and reducing the total 2.4 Dexterity metric number of vertices. Subsequently, the polygon models within To evaluate the positioning and orientation capability of the the series are interpolated by morphing and spatial alignment tool tip at all given target points, we defined a scoring metric. from which the target workspace volume is generated. The The orientation around the spatial axes is not continuously alignment can be application specific, e.g., for horizontal or analysed but tested in discrete steps [10], [1]. In these steps, vertical patient positions. Alternatively, in the target the angle of attack of the instrument is varied with respect to workspace model, irrelevant body regions can be removed, the phantom surface. Not all axes need to be varied if the e.g., the extremities. instrument is inherently rotatable. Kinematic simulation scripts were developed using the Fig. 1b shows a five-stage discretization scheme in which Virtual Robot Experimentation Platform V-REP (Coppelia at first the virtual tool is set to the same orientation as the target Robotics GmbH, Zurich Switzerland). Its integrated inverse point vertex and is subsequently varied by +30° and -30° for kinematics IK Calculation Module was used. The open motion the pitch and roll axis of the instrument. The score is increased planning library plug-in [12] was used to generate collision- by one for each collision-free position reached. The maximum free paths. achievable score value per single target point in this scheme is therefore 5. In our results, we used this scoring range in Fig. 3. 3 Results 2.4.1 Software and simulation framework We simulated a series with a target workspace phantom using the aforementioned properties for the simulation of the The three-dimensional body models are created using the ultrasound imaging task with a lying patient phantom and four UMTRI Human Shape online platform (University of different kinematics (Fig. 2a). The first kinematics (virtual Michigan, USA), which is mainly used for ergonomics studies model of a 7-joint robot KUKA LBR4 +) served as a reference in product design. This tool contains a statistical and was replaced by three optimized kinematics (the 6-, 7- anthropometric data model prepared from whole-body laser and 8-joint kinematics). The spatial arrangement of the robot scans of humans. After selection of physiognomic parameters, base with respect to the patient model was placed at a distance a series of two or more differently dimensioned polygonal of 50 cm from the centre of the torso at the height level of the models of the body surface can be exported by this tool into lying surface. The design of the three self-made kinematics structured text files. with associated parameters is shown in Fig. 2b. In the design These polygon models are modified by custom scripts of the kinematic basic structure, only rotational joints with using 3D Blender Suite (Blender Foudation, Amsterdam, The S. Böttger et al., Medical robotics simulation framew ork for application-specific optimal kinematics — 148 potentially be decreased to an acceptable level. In this study we investigated the benefit of optimized kinematic structures in medical applications. This is the first step towards a fully automated framework that optimizes the robot placement and the tool geometry in addition to kinematic structures. In future work we will use such a system also for other applications beyond medical robotics. Author Statement Research funding: The author state no funding involved. Conflict of interest: Authors state no conflict of interest. Figure 4: (a) Graphical visualization of the spatial dexterity Informed consent: Informed consent has been obtained from distribution for the reference kinematics (a) and the self- all individuals included in this study. Ethical approval: Not designed 6-, 7- and 8-joint kinematics (b, c, d) in the target applicable. space. The same colors as in Fig. 3 were used to indicate the dexterity score value on target points. axial offset as well as from distal to proximal increasing link References lengths were selected. [1] Yoshikawa T. Manipulability and redundancy control of We compared the mean score value of all end effector robotic mechanisms. Proceedings IEEE International positions of our baseline, the KUKA LBR4+, 2.1 ± 2.1 (mean Conference on Robotics and Automation vol. 2 1985:1004–1009. ± standard deviation), with optimized structures of a 6-joint [2] Patel S, Sobh T. Manipulator Performance Measures - A arm 3.4 ± 2.0, a 7-joint arm 3.5 ± 2.0 and an 8-joint arm 4.1 ± Comprehensive Literature Survey. J. Intell. Robot. Syst. Theory Appl. vol. 77 no. 3–4 2014:547–570. 1.6. Fig. 3 shows the numerical simulation results for the four [3] Paden B, Sastry S, Optimal Kinematic Design of 6R examined kinematics. Manipulators. Int. J. Rob. Res. vol. 7 no. 2 1988:43–61. The spatial dexterity distribution of the non-optimized [4] Nelson CA, Laribi MA, Zeghloul S. Optimization of a Redundant Serial Spherical Mechanism for Robotic kinematics focuses more on a lateral region in target space Minimally Invasive Surgery. In: Zeghloul S, Romdhane L, while it spreads more evenly with optimized kinematics and Laribi M. editors. Computational Kinematics. Mechanisms and Machine Science vol 50. Springer. Cham: 2018:126– an increasing number of joints as shown in Fig. 4. [5] Pamanes GJA, Zeghloul S. Optimal placement of robotic manipulators using multiple kinematic criteria. Proceedings IEEE International Conference on Robotics and Automation 2002:933–938. 4 Conclusion [6] Zeghloul S, Pamanes-Garcia JA. Multi-criteria optimal placement of robots in constrained environments In: A simulation framework was developed to investigate an Robotica vol. 11 no. 2 1993:105–110. [7] Vidaković J, Jerbić B, Švaco M, Šuligoj F, Šekoranja B. objective assessment of kinematics for their usability in Position planning for collaborating robots and its medical robotic applications. Due to the complex relationship application in neurosurgery In: Teh. Vjesn. - Tech. Gaz., vol. 24 no. 6 2017. between kinematic design and resulting dexterity, such [8] Xiang X. Simulation and Analysis of Three-Dimensional examinations cannot be intuitively tested, but require Space Path Prediction for Six-Degree-of-Freedom (SDOF) Manipulator In: 3D Res. vol. 10 no. 2 2019:15. systematic evaluation. In the experiment, a commercial high [9] Siciliano B, Khatib O. Springer Handbook of Robotics. dexterity robot was used as a reference. We compared a 6-joint Berlin, Heidelberg: Springer Berlin Heidelberg; 2008. kinematics to redundant designs and we found that redundancy [10] Vijaykumar R, Waldron KJ, Tsai MJ. Geometric Optimization of Serial Chain Manipulator Structures for does not necessarily lead to a high dexterity, and can be Working Volume and Dexterity. Int. J. Rob. Res. vol. 5 no. compensated or outperformed by structural optimization. This 2 1986:91–103. [12] Sucan IA, Moll M, Kavraki LE. The Open Motion Planning includes a larger arm length and joint displacement enabling Library. IEEE Robot. Autom. Mag. vol. 19 no. 4 2012:72– over-rotation of the joints. The implementation of structural improvements is more cost effective in real hardware than [13] Nouaille L, Laribi M, Nelson C, Essomba T, Poisson G, Zeghloul S, Design process for robotic medical tool appending additional joints. guidance manipulators. Proc. Inst. Mech. Eng. Part C J. Ultimately, by using lightweight and simple medical Mech. Eng. Sci. vol. 230 no. 2 2016:259–275 robots with a correspondingly low energy balance and a specific design that satisfy the requirements, the injury risk can http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Current Directions in Biomedical Engineering de Gruyter

Medical robotics simulation framework for application-specific optimal kinematics

Loading next page...
 
/lp/de-gruyter/medical-robotics-simulation-framework-for-application-specific-optimal-Qgm1T1yf8b

References (11)

Publisher
de Gruyter
Copyright
© 2019 by Walter de Gruyter Berlin/Boston
eISSN
2364-5504
DOI
10.1515/cdbme-2019-0037
Publisher site
See Article on Publisher Site

Abstract

Current Directions in Biomedical Engineering 2019;5(1):145-148 Sven Böttger*, Tolga-Can Çallar, Achim Schweikard and Elmar Rueckert Medical robotics simulation framework for application-specific optimal kinematics Abstract: Most kinematic structures in robot architectures Keywords: medical robotics, robot kinematics, optimization, for medical tasks are not optimal. Further, the workspace anthropometric body shape data and payloads are often oversized which results in high https://doi.org/10.1515/cdbme-2019-0037 product prices that are not suitable for a clinical technology transfer. To investigate optimal kinematic structures and configurations, we have developed an adaptive simulation framework with an associated workflow for requirement 1 Introduction analyses, modelling and simulation of specific robot kinematics. The framework is used to build simple and cost For medical robotic applications, standard industrial robots are effective medical robot designs and was evaluated in a tool often used because of their good commercial availability, manipulation task where medical instruments had to be product quality, and accuracy, although they do not optimally positioned precisely and oriented on the patient's body. The meet the kinematic requirements for the application. These model quality is measured based on the maximum systems are designed for universal use for a variety of tasks in workspace coverage according to a configurable scoring various industries and are therefore usually oversized in metric. The metric generalizes among different human workspace and payload, whilst also being expensive and body shapes that are based on anthropometric data from requiring special security measures. UMTRI Human Shape. This dexterity measure is used to Some related studies that investigate optimal kinematic analyze different kinematic structures in simulations using structures in medical robotics exist. Yoshikawa [1] discussed the open source simulation tool V-REP. Therefor we the manipulating ability of robotic mechanisms in positioning developed simulation and visualization procedures for and orienting end-effectors and proposes a measure of medical tasks based on a patchwork of size-variant manipulability. Some performance measures were reviewed anatomical target regions that can be configured and by Patel [2]. Paden [3] formulated an optimality theorem for selectively activated in a motion planning controller. In our six revolute joints kinematics and Nelson [4] proposed a evaluations we compared the dexterity scores of a Monte Carlo simulation algorithm for optimizing a redundant commercial lightweight robot arm with 7 joints to serial spherical linkage. Pamanes [5], Zeghloul [6] and optimized kinematic structures with 6, 7 and 8 joints. Com- Vidaković [7] presented methods to find the optimal pared to the commercial hardware, we achieved placement of robots. Xiang [8] proposes a three-dimensional improvements of 59% when using an optimized 6- space path prediction simulation method and a design process dimensional robot arm, 64% with the 7-dimensional arm method for robotic medical tool-guidance manipulators was and 96% with an 8-dimensional robot arm. Our results proposed by Nouaille [13]. show that simpler robot designs can outperform the In this work, we perform the analysis of the kinetic typically used commercial robot arms in medical requirements, in particular of workspace and dexterity, applications where the maximum workspace coverage is covering all the above issues. Furthermore, we determine essential. Our framework provides the basis for a fully design and a configuration of an optimal robot, especially for automatic optimization tool of the robot parameters that can applications in the medical field for the manipulation of be applied to a large variety of problems. instruments on the human body, e.g. robot-assisted ultrasound or needle puncture. ______ *Corresponding author: Sven Böttger: e-mail: 2 Methods boettger@rob.uni-luebeck.de Tolga-Can Çallar, Achim Schweikard, Elmar Rückert: Institute for Robotics and Cognitive To analyse, model and simulate the application-specific Systems, Universität zu Lübeck, Ratzeburger Allee 160, Lübeck, optimal robot kinematics we developed and evaluated a Germany Open Access. © 2019 Sven Böttger et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0 License. S. Böttger et al., Medical robotics simulation framework for application -specific optimal kinematics — 146 workflow and a software framework. The application-specific range and axial offset of the joints and the arm link lengths) target workspaces were modelled and compared with the robot and the design of the tool geometry of the application-specific workspaces, and the score values were determined according instrument given by the tool-tip coordinates. The general to a dexterity metric, and hence, a manual optimization was functional requirements for the kinematic design were carried out. evaluated using the design process by Siciliano [9]. The design goal is a simple, lightweight, dynamically stable and cost- effective robotic arm. The structure with as few joints as 2.1 Computing targets using anatomical possible represents the most important optimization criterion that meets all goal criteria. body models The first section of the workflow was the requirement analysis. We utilized a statistically representative anatomical 2.3 Dexterity simulation body model based on anthropometric data containing some In the simulation, as the third workflow section, the two model variations regarding size and shape to cover the spectrum of components were integrated into a common simulation human anatomy (section Software and simulation framework). environment and virtually interconnected, whereby the kinetic The target areas for the intended medical applications were performance was determined. A test algorithm has been marked by selecting vertices from the surface of the body implemented to approximate the functional requirements of model. Likewise, different regions can be combined with each the intended medical application and tests the kinematics using other depending on the application scenario. a dexterity metric defined as follows: It tests whether the In a second workflow step, the models for the target instrument could be virtually positioned at all target positions workspace and the robot workspace were modelled in parallel. and oriented in some different rotations (next section) in The target workspace was created by generating multiple body parameterized Tait–Bryan (roll-pitch-yaw) angles. Target model variations by adjusting the physiognomic parameters positions for testing are all selected vertex points contained in for gender, patient size, body mass index, body length, and age all layers of the target workspace model. Those were assigned to produce a minimal and maximal shell model of the body with direction vectors normal to the model surface. The base surface. Based on the DINED Anthropometric Database [1], th of the robot model was positioned relative to the target model this range was selected from the 5 percentile (corresponding in the simulation space, whereby the position has significant to the body height of 1.54 m with a body mass index [BMI] of th impact on the result and is also subject to optimization. The 20) to 95 percentile (1.91 m, BMI = 34) of the anatomical dexterity estimation was implemented using an algorithm for bandwidth (see Fig. 1a). inverse kinematics calculation and path planning (section The generated model data follows an identical general Software and simulation framework), which attempts to find vertex structure, only differing in spatial deformation all target configurations on collision-free paths from the according to the input parameters. Subsequently, the mapping starting position to the target position. Self-collisions and of the application-specific target areas to the likewise (layer-selective) collisions with the phantom must be avoided. application-specifically dimensioned shell models is performed by assigning the vertex indices. In the following morphing between minimum and maximum shell any number of intermediate shells can be created, controlling the accuracy of the dexterity calculation. The superposition of all vertex points belonging to the target areas from all shells yields the spatial target workspace. Finally, the number of vertex points is reduced to an appropriate ratio (in the present experiment to 1533) in order to reduce the computational demands in the th th Figure 1: (a) Anthropometric shell model generation of 5 and 95 simulation. percentile of the population and an intermediate shell by morphing operations (1), marking of grouped colored target areas (2), cropping (3) and superposition in supine (4, frontal 2.2 Optimizing kinematic structures and lateral view) yields the 3D target workspace embracing solely the colored vertices. (b) Discretization scheme of 5 orientations of the medical instrument for dexterity test to apply on every vertex position in target space relative to their We varied all the relevant kinematic parameters that form the direction. workspace of an articulated robot arm (namely number, types, S. Böttger et al., Medical robotics simulation framework for applicatio-nspecific optimal kinematics — 147 Figure 2: (a) Virtual simulation scene including target workspace phantom and 6-Joint kinematic. (b) Design and dimensions of the self-designed kinematic structures.. For planning of the collision-free paths to the target configurations, a sample based tree planner search algorithm (section Software and simulation framework) operating in Figure 3: Numerical dexterity distribution of the four examined joint parameter space is used. It selects always the shortest robot kinematics including mean dexterity score and standard path out of the generated plans. deviation. In each experiment 1533 targets had to be reached. Netherlands), highlighting the vertex groups belonging to the target areas within the polygon models and reducing the total 2.4 Dexterity metric number of vertices. Subsequently, the polygon models within To evaluate the positioning and orientation capability of the the series are interpolated by morphing and spatial alignment tool tip at all given target points, we defined a scoring metric. from which the target workspace volume is generated. The The orientation around the spatial axes is not continuously alignment can be application specific, e.g., for horizontal or analysed but tested in discrete steps [10], [1]. In these steps, vertical patient positions. Alternatively, in the target the angle of attack of the instrument is varied with respect to workspace model, irrelevant body regions can be removed, the phantom surface. Not all axes need to be varied if the e.g., the extremities. instrument is inherently rotatable. Kinematic simulation scripts were developed using the Fig. 1b shows a five-stage discretization scheme in which Virtual Robot Experimentation Platform V-REP (Coppelia at first the virtual tool is set to the same orientation as the target Robotics GmbH, Zurich Switzerland). Its integrated inverse point vertex and is subsequently varied by +30° and -30° for kinematics IK Calculation Module was used. The open motion the pitch and roll axis of the instrument. The score is increased planning library plug-in [12] was used to generate collision- by one for each collision-free position reached. The maximum free paths. achievable score value per single target point in this scheme is therefore 5. In our results, we used this scoring range in Fig. 3. 3 Results 2.4.1 Software and simulation framework We simulated a series with a target workspace phantom using the aforementioned properties for the simulation of the The three-dimensional body models are created using the ultrasound imaging task with a lying patient phantom and four UMTRI Human Shape online platform (University of different kinematics (Fig. 2a). The first kinematics (virtual Michigan, USA), which is mainly used for ergonomics studies model of a 7-joint robot KUKA LBR4 +) served as a reference in product design. This tool contains a statistical and was replaced by three optimized kinematics (the 6-, 7- anthropometric data model prepared from whole-body laser and 8-joint kinematics). The spatial arrangement of the robot scans of humans. After selection of physiognomic parameters, base with respect to the patient model was placed at a distance a series of two or more differently dimensioned polygonal of 50 cm from the centre of the torso at the height level of the models of the body surface can be exported by this tool into lying surface. The design of the three self-made kinematics structured text files. with associated parameters is shown in Fig. 2b. In the design These polygon models are modified by custom scripts of the kinematic basic structure, only rotational joints with using 3D Blender Suite (Blender Foudation, Amsterdam, The S. Böttger et al., Medical robotics simulation framew ork for application-specific optimal kinematics — 148 potentially be decreased to an acceptable level. In this study we investigated the benefit of optimized kinematic structures in medical applications. This is the first step towards a fully automated framework that optimizes the robot placement and the tool geometry in addition to kinematic structures. In future work we will use such a system also for other applications beyond medical robotics. Author Statement Research funding: The author state no funding involved. Conflict of interest: Authors state no conflict of interest. Figure 4: (a) Graphical visualization of the spatial dexterity Informed consent: Informed consent has been obtained from distribution for the reference kinematics (a) and the self- all individuals included in this study. Ethical approval: Not designed 6-, 7- and 8-joint kinematics (b, c, d) in the target applicable. space. The same colors as in Fig. 3 were used to indicate the dexterity score value on target points. axial offset as well as from distal to proximal increasing link References lengths were selected. [1] Yoshikawa T. Manipulability and redundancy control of We compared the mean score value of all end effector robotic mechanisms. Proceedings IEEE International positions of our baseline, the KUKA LBR4+, 2.1 ± 2.1 (mean Conference on Robotics and Automation vol. 2 1985:1004–1009. ± standard deviation), with optimized structures of a 6-joint [2] Patel S, Sobh T. Manipulator Performance Measures - A arm 3.4 ± 2.0, a 7-joint arm 3.5 ± 2.0 and an 8-joint arm 4.1 ± Comprehensive Literature Survey. J. Intell. Robot. Syst. Theory Appl. vol. 77 no. 3–4 2014:547–570. 1.6. Fig. 3 shows the numerical simulation results for the four [3] Paden B, Sastry S, Optimal Kinematic Design of 6R examined kinematics. Manipulators. Int. J. Rob. Res. vol. 7 no. 2 1988:43–61. The spatial dexterity distribution of the non-optimized [4] Nelson CA, Laribi MA, Zeghloul S. Optimization of a Redundant Serial Spherical Mechanism for Robotic kinematics focuses more on a lateral region in target space Minimally Invasive Surgery. In: Zeghloul S, Romdhane L, while it spreads more evenly with optimized kinematics and Laribi M. editors. Computational Kinematics. Mechanisms and Machine Science vol 50. Springer. Cham: 2018:126– an increasing number of joints as shown in Fig. 4. [5] Pamanes GJA, Zeghloul S. Optimal placement of robotic manipulators using multiple kinematic criteria. Proceedings IEEE International Conference on Robotics and Automation 2002:933–938. 4 Conclusion [6] Zeghloul S, Pamanes-Garcia JA. Multi-criteria optimal placement of robots in constrained environments In: A simulation framework was developed to investigate an Robotica vol. 11 no. 2 1993:105–110. [7] Vidaković J, Jerbić B, Švaco M, Šuligoj F, Šekoranja B. objective assessment of kinematics for their usability in Position planning for collaborating robots and its medical robotic applications. Due to the complex relationship application in neurosurgery In: Teh. Vjesn. - Tech. Gaz., vol. 24 no. 6 2017. between kinematic design and resulting dexterity, such [8] Xiang X. Simulation and Analysis of Three-Dimensional examinations cannot be intuitively tested, but require Space Path Prediction for Six-Degree-of-Freedom (SDOF) Manipulator In: 3D Res. vol. 10 no. 2 2019:15. systematic evaluation. In the experiment, a commercial high [9] Siciliano B, Khatib O. Springer Handbook of Robotics. dexterity robot was used as a reference. We compared a 6-joint Berlin, Heidelberg: Springer Berlin Heidelberg; 2008. kinematics to redundant designs and we found that redundancy [10] Vijaykumar R, Waldron KJ, Tsai MJ. Geometric Optimization of Serial Chain Manipulator Structures for does not necessarily lead to a high dexterity, and can be Working Volume and Dexterity. Int. J. Rob. Res. vol. 5 no. compensated or outperformed by structural optimization. This 2 1986:91–103. [12] Sucan IA, Moll M, Kavraki LE. The Open Motion Planning includes a larger arm length and joint displacement enabling Library. IEEE Robot. Autom. Mag. vol. 19 no. 4 2012:72– over-rotation of the joints. The implementation of structural improvements is more cost effective in real hardware than [13] Nouaille L, Laribi M, Nelson C, Essomba T, Poisson G, Zeghloul S, Design process for robotic medical tool appending additional joints. guidance manipulators. Proc. Inst. Mech. Eng. Part C J. Ultimately, by using lightweight and simple medical Mech. Eng. Sci. vol. 230 no. 2 2016:259–275 robots with a correspondingly low energy balance and a specific design that satisfy the requirements, the injury risk can

Journal

Current Directions in Biomedical Engineeringde Gruyter

Published: Sep 1, 2019

Keywords: medical robotics; robot kinematics; optimization; anthropometric body shape data

There are no references for this article.