Clarke Transform and Encoder-Decoder Architecture for Arbitrary Joints Locations in Displacement-Actuated Continuum Robots

Reinhard M. Grassmann, Jessica Burgner-Kahrs · 2024

Read Clarke Transform and Encoder-Decoder Architecture for Arbitrary Joints Locations in Displacement-Actuated Continuum Robots by Reinhard M. Grassmann, Jessica Burgner-Kahrs (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-16401.txt.

Introduction

Reinhard Grassmann and Jessica Burgner-Kahrs extend the Clarke transform for continuum robots in work from 2024. These soft robots bend along a center line and are driven by displacements at joints that may sit anywhere along it.

Past formulas assumed symmetrically placed joints, which limited design freedom. The revised transform handles any number of joints at any positions. An encoder-decoder built on it converts joint values between different robot designs, enabling direct knowledge transfer. Control and path generation demos in simulation show the method plugs into existing frameworks built for the symmetric case.

Catalog record

CREATORReinhard M. Grassmann, Jessica Burgner-Kahrs
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-16401

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Clarke Transform and Encoder-Decoder Architecture for Arbitrary Joints Locations in Displacement-Actuated Continuum Robots Reinhard M. Grassmann, Jessica Burgner-Kahrs Abstract In this paper, we consider an arbitrary number of joints and their arbitrary joint locations along the center line of a displacement-actuated continuum robot. To achieve this, we revisit the derivation of the Clarke transform leading to a formulation capable of considering arbitrary joint locations. The proposed modified Clarke transform opens new opportunities in mechanical design and algorithmic approaches beyond the current limiting dependency on symmetric arranged joint locations. By presenting an encoder-decoder architecture based on the Clarke transform, joint values between different robot designs can be transformed enabling the use of an analogous robot design and direct knowledge transfer. To demonstrate its versatility, applications of control and trajectory generation in simulation are presented, which can be easily integrated into an existing framework designed, for instance, for three symmetric arranged joints. ©2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. I Introduction Real-world tasks in medical and industrial applications do not impose symmetric manipulator designs for soft and continuum robots. However, almost all displacement-actuated continuum robots rely on a very narrow design space, i.e. , three or four symmetric arranged joints. Overcoming that design restriction has obvious benefits. In particular, increasing the joint number n significantly, i.e. , n\!\gg\!3 , and considering asymmetric joint arrangements: (i) enhances manipulability along the directions that coincide with joint locations and center-line; (ii) improves force absorption and delivery due to the distributed nature of the actuation forces; and (iii) increases safety through actuation redundancy. Desirable in medical applications [ 1 , 2 ] and industrial settings [ 3 , 4 ] , the utilization of such mechanical design may lead to increased load capacity, better shape conformation, enhanced stability, and variable stiffness. Therefore, exploring and providing a computational inexpensive general approach is a pioneering step towards more capable displacement-actuated continuum robots that include a large set of continuum and soft robots. To this day, displacement-actuated continuum robots with n joints are under-explored. Some attempts, e.g. , [ 5 , 6 ] for n joints provide the solution for the robot-independent mapping, and the work by Dalvand et al. [ 7 ] considers n joints in their forward kinematics framework. Their framework includes an exhaustive list of 2^{n} combinations of passive and active tendons as joints and several branches, e.g. , if-else-statements, as well as function calls to numerically solve an underlying beam model. However, a computed result might not be consistent, where a consistency between 96.1\text{\,}\% and 99.2\text{\,}\% is reported. While this approach [ 7 ] can be considered to account for arbitrary joint locations, it relies heavily on computing and heuristics. [ ] [ ] [ ] [ ] Fig. 1: Joint location and improved joint representation. The kinematics of a displacement-actuated continuum robot with fixed segment length is mainly influenced by the number of joints n and their location in the cross-section in terms of polar coordinates. For the i\textsuperscript{th} joint, the polar coordinates is described by the distance d_{i} to the center-line and the angle \psi_{i} . Joint representations, e.g. , [ 6 , 8 , 9 , 10 ] , have been proposed to consider various arrangement. Our approach generalizes and covers all cases. In our recent work [ 9 ] , they show that the Clarke transform using a generalized Clarke transformation matrix can be used to map n joint values onto a two-dimensional manifold. Kinematics based on the Clarke transform are branchless, closed-form, and singularity-free. However, this approach [ 9 ] only applies to symmetric arranged joint location, see Fig. 1 . The Clarke transform can also account for non-constant distances [ 10 ] . However, to the best of our knowledge, no approach can consider arbitrary joint location while only considering the kinematic design parameters of the target displacement-actuated continuum robot. The kinematic design parameters entail angular offset \psi_{i} , distance d_{i} to center-line in the cross-section, and length l of a displacement-actuated continuum robot as illustrated in Fig. 2 . Due to the Clarke transform’s relationship to the Clarke transformation matrix, looking into generalized Clarke transformation matrices for n phases in the literature for electrical motors is worthwhile. Furthermore, based on the analogy to Kirchhoff’s current law [ 9 ] , the actuation constraint, i.e. , \sum_{i=1}^{n}\rho_{i}=0 , where \rho_{i} are the displacement values, inherent to a displacement-actuated continuum robot with symmetric arrangement can be considered as balanced system. Consequently, a displacement-actuated continuum robot with asymmetric arrangement can be considered an unbalanced system as the constraint is not necessarily zero akin to an unbalanced electrical system, where the sum of all electrical current in each phase is non-zero. Janaszek [ 11 ] derives a similar matrix to ours [ 10 , 9 ] , which differs only by a scaling factor. In work by Willems [ 12 ] and by Rockhill & Lipo [ 13 ] , the squared matrix is only useful for balanced systems. For symmetric phase arrangements, all variants should produce \left(n-2\right) zeros resulting in unnecessary computation and large state representation. Another squared matrix is derived by Willems [ 12 ] , which can consider unbalanced systems. However, for all representations derived by Willems [ 12 ] , it is not guaranteed that its inverse produces always \left(n-2\right) zeros. Therefore, the dimensionality cannot be reduced to two variables. To summarize, none of the generalized Clarke transformation matrices can be used directly. In the present work, we propose a modified generalized Clarke transformation matrix to construct a Clarke transform that applies to arbitrary joint locations for displacement-actuated continuum robots. Furthermore, we propose an encoder-decoder architecture based on the Clarke transform to facilitate the use of those continuum robots. In particular, the contribution of this paper includes: • Deriving a generalized Clarke transformation matrix for unbalanced systems • Modifying the Clarke transform to consider arbitrary joint locations • Proposing an encoder-decoder architecture to transform joint values between different robot designs • Adapting a \mathcal{C}^{4} -smooth trajectory generator • Generating feasible joint values Due to the use of Clarke coordinates, this approach automatically contributes to providing robot-dependent mapping. Furthermore, all derived approaches are linear, compact, and closed-form. II Extending Clarke Transform In this section, we briefly revisit the Clarke transform and state the relationship to arc parameters. Afterward, an alternative derivation is stated that does not rely on the connection to a Clarke transformation matrix. Based on the alternative derivation, we derive a Clarke transform and its inverse to consider asymmetric arranged joint locations. II-A Clarke Transform for Displacement-Actuated Joint Using the representation proposed in [ 9 ] for i\textsuperscript{th} entry of displacement-actuated joints \boldsymbol{\rho} given by \rho_{i} , the index notation for \boldsymbol{\rho} is \displaystyle\boldsymbol{\rho}=\left(\rho_{\text{Re}}\cos{\psi_{i}}+\rho_{\text{Im}}\sin{\psi_{i}}\right)_{i}\subset\mathbb{R}^{n}, (1) whereas Clarke coordinates, i.e. , \rho_{\mathrm{Re}} and \rho_{\mathrm{Im}} , being the free parameter of ( 1 ), can be combined into \displaystyle\overline{\boldsymbol{\rho}}=\left[\rho_{\mathrm{Re}},\rho_{\mathrm{Im}}\right]^{\top}\in\mathbb{R}^{2}. (2) Note that ( 1 ) is an element

Read the complete text