Discrete Kinematic Modeling for Soft Continuum Robots

35 concepts32 papers1 proposal

Soft robots and continuum manipulators—inspired by the flexible dexterity of octopus tentacles and elephant trunks—possess theoretically infinite degrees of freedom. While this flexibility allows them to navigate cramped environments and interact safely with humans, it creates a "mathematical nightmare" for real-time control and path planning. To solve this, researchers have developed a suite of modeling techniques that approximate continuous deformation using discrete, computationally efficient frameworks.

The Constant Curvature Approximation

The most common strategy for simplifying continuum kinematics is the Piecewise Constant Curvature (PCC) model. Instead of treating a flexible segment as a complex, unpredictable curve, the authors of several studies, including McMahan et al. and Jiang et al., propose approximating these segments as circular arcs.

By assuming each segment has a constant curvature, the robot's infinite-dimensional state is reduced to a finite set of geometric parameters. This allows researchers to adapt traditional robotics frameworks, such as Denavit-Hartenberg parameters, to soft structures. For example, Arachchige et al. utilized a floating-base kinematic framework that treats limb bending as circular arcs to map actuator changes to task-space coordinates, enabling precise locomotion in modular soft quadrupeds. Similarly, Ghoreishi et al. applied this assumption to multi-segmented robotic catheters, using Euler-Bernoulli beam theory to derive analytical equations that are fast enough for real-time surgical navigation.

While PCC models are geometrically intuitive, they can suffer from numerical instabilities, particularly when curvature approaches zero. To address this, Cheng et al. introduced the APCC 2L-5R model, which replaces a continuous segment with a discrete "2-link, 5-joint" rigid equivalent. By mapping the kinematics of a continuum robot onto a rigid-body framework, the authors avoid the singularities common in traditional models, ensuring stable real-time shape estimation.

This discrete approach is also effective for modeling thin materials. Kubota et al. demonstrated that a deformable plastic tape could be modeled as a kinematic chain of rigid rectangular plates. By optimizing the angles between these plates to match visual silhouettes, the system achieves the speed necessary for robotic handling on assembly lines.

Physics-Informed Rod Theories

For applications requiring higher physical fidelity—such as simulating twisting, shearing, or environmental contact—researchers often turn to Cosserat rod theory and Discrete Elastic Rods (DER). Unlike simple geometric arcs, these models account for the internal forces and material properties of the robot.

Wang et al. highlight that Cosserat rod theory provides a structured mathematical framework for representing slender structures by tracking a centerline and its associated rotations. This ensures that reconstructed shapes remain kinematically consistent with real-world material behavior. Naughton et al. implemented this in the Elastica simulation environment, allowing reinforcement learning agents to exploit "mechanical intelligence" by leaning against obstacles to solve tasks.

For micro-robotics, Du et al. employed a DER framework that discretizes a robot into nodes and edges. This allows for the simultaneous simulation of elastic forces (stretching and bending) and external hydrodynamic forces, which is critical for robots navigating fluid interfaces.

Complexity Reduction in Planning and Control

The ultimate goal of these discrete approximations is to decouple planning complexity from the robot's physical degrees of freedom. Hayashi et al. showed that by treating a highly redundant manipulator as a continuous curve during the initial pathfinding phase, the computational cost of obstacle avoidance becomes polynomial rather than exponential.

Modern frameworks continue to build on this foundation. Luo et al. integrated PCC modeling with RRT* algorithms, accounting for the mechanical coupling between segments to ensure safety in dynamic environments. By reducing the robot's state to a manageable set of variables, these models enable advanced control strategies—from the "Fuzzy Backstepping" controllers proposed by Jalali et al. to the Vision-Language-Action models deployed by Meo et al.—to function effectively on compliant hardware.

Key Foundational Work

3D Shape and Pose Estimaion of Deformable Tapes from Multiple Views (2004) — contributed 1 concept elements to this direction.
Path Planning for Highly Redundant Manipulators Using a Continuous Model (1991) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_piecewise_constant_curvature_modeling_436a6177,
  title = {piecewise constant curvature modeling},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/piecewise-constant-curvature-modeling/rdr_436a617715bb45d98da495734e34661b},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-21},
}

This exact version (version 1, frozen text):

@misc{lacuna_piecewise_constant_curvature_modeling_436a6177_v1,
  title = {piecewise constant curvature modeling},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/piecewise-constant-curvature-modeling/rdr_436a617715bb45d98da495734e34661b/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-21},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake