Morphing of and writing with a scissor linkage mechanism

Mohanraj A; S Ganga Prasath

cs.ROFeb 16, 2026

Morphing of and writing with a scissor linkage mechanism

Authors: Mohanraj A, S Ganga Prasath

Organizations: Department of Applied Mechanics & Biomedical Engineering, IIT Madras, Chennai, TN 600036.

Abstract

Kinematics of mechanisms is intricately coupled to their geometry and their utility often arises out of the ability to perform reproducible motion with fewer actuating degrees of freedom. In this article, we explore the assembly of scissor-units, each made of two rigid linear members connected by a pin joint. The assembly has a single degree of freedom, where actuating any single unit results in a shape change of the entire assembly. We derive expressions for the effective curvature of the unit and the trajectory of the mechanism's tip as a function of the geometric variables which we then use as the basis to program two tasks in the mechanism: shape morphing and writing. By phrasing these tasks as optimization problems and utilizing the differentiable simulation framework, we arrive at solutions that are then tested in table-top experiments. Our results show that the geometry of scissor assemblies can be leveraged for automated navigation and inspection in complex domains, in light of the optimization framework. However, we highlight that the challenges associated with rapid programming and error-free implementation in experiments without feedback still remain.

Figures & tables

Figure 1: Geometry of scissor linkage mechanism. (A) A single scissor-unit is made of two rigid members of length $l$ connected by a pin joint. Each unit is described by two geometric variables: angle between members $\phi$ ; the aspect ratio $\alpha$ defined as relative distance of the pin joint from the end to the total length, $l$ . (B) Vectors $\hat{\mathbf{t}}^{(1)}$ and $\hat{\mathbf{t}}^{(2)}$ capture the orientation of the members $ac$ , $bd$ . Along the edges $ac$ , $bd$ we define two normal vectors $\hat{\mathbf{N}}_{e}$ and $\hat{\mathbf{N}}_{w}$ . The tangent vectors, $\hat{\mathbf{T}}_{e}$ and $\hat{\mathbf{T}}_{w}$ , are defined orthogonal to these normal vectors. $\Delta_{o}$ is the width of the segment passing through the pin joint and intersecting faces $ad,bc$ . (C) Contours of $\kappa_{o}(\alpha,\phi)$ , the effective curvature of each scissor-unit defined in Eq. 2 . $\kappa_{o}$ diverges for $\phi\to 0$ and vanishes at the symmetric midpoint $\alpha=0.5$ . (D) Comparison between theoretical prediction of the critical actuation angle $\Psi^{*}$ (solid line) required to achieve self-intersection ( $N\phi^{*}=2\pi$ ) and experimentally measured angles (green markers). Error bars are smaller than the size of the markers. (E) Scissor-units are assembled by connecting to neighboring units via pin joints resulting in a single degree-of-freedom for the mechanism. The location of center pin joint of $j$ -th unit is denoted by $\mathbf{r}_{j}$ and the radius of curvature by $\kappa_{j}^{-1}$ . Actuation angle $\Psi$ is the angle between the members of the first unit, $\Psi=\phi_{1}$ and is used to control the motion of the assembly. $\phi^{*}$ is defined as the angle between the lines connecting the sides $ad$ and $bc$ in Fig. 1 (B)). (F) Sequence of configurations of a scissor mechanism from experiments whose aspect ratios are given by $\{\alpha_{j}\}=0.45$ for $0<j<6$ and $\{\alpha_{j}\}=0.55$ for $6\leq j<12$ . Decreasing the actuation angle $\Psi$ via Arduino controlled motor drive (see Sec. II for details) steers the mechanism from a near-straight configuration (left) to a curved shape (right), illustrating large-scale shape change enabled by the intrinsic geometry.

Figure 2: Shape morphing of scissor mechanism. (A) 3D printed scissor-units made with an adjustable slider allowing continuous variation in the aspect ratio

\alpha

along with a screw that can be locked into position. These units are connected to neighboring units via pin joints. (B) In the shape morphing task, the mechanism is designed to take a target shape by choosing appropriate aspect ratios

\{\alpha_{j}\}

and actuating the mechanism to angle

\Psi^{*}

. These values are arrived at by minimizing the shape error,

\mathcal{L}

between the current configuration,

\mathbf{r}_{j}

and the target shape described by a curvature

\kappa^{t}

(see SI Sec. S5 for details). (C) Examples showing the scissor mechanism deployed to target shapes (curves in blue) in experiments and simulation (see SI Video 2 for the deployment sequence). The target shapes have

(i)

a monotonically increasing curvature (left);

(ii)

a sinusoidal curvature profile (middle);

(iii)

a three-petaled flower (right). (D) Evolution of the mechanism shape in experiments upon actuation by decreasing

\Psi

. Inset shows the solution aspect ratios

\alpha_{j}

obtained via the optimization procedure detailed in Sec. III.3 .

Figure 3: Writing with a scissor mechanism. (A) Contour in the $(\alpha,\Psi)$ plane showing the curvature of the distal end of the mechanism $\kappa_{\textrm{tip}}(\alpha,\Psi;N)$ made of $N$ scissor-units with constant $\alpha$ . Three different contours corresponding to $N=5$ (orange), $N=10$ (green), and $N=20$ (blue) are shown and iso-contour labels correspond to magnitude of $\kappa_{\textrm{tip}}$ . Insets illustrate tip trajectories $\mathbf{r}_{\textrm{tip}}(\Psi)$ (for $N=5$ ) obtained by sweeping $\Psi$ from $\pi$ to $0$ at the corresponding parameter locations. As $\alpha$ departs from $0.5$ the trajectory rapidly develops tight loops (large $\kappa_{\textrm{tip}}$ ), whereas $\alpha\to 0.5$ yields an almost straight path with $\kappa_{\textrm{tip}}\approx 0$ . We see that the mechanism is capable of tracing trajectories with rapidly changing tip curvature, however, they vary only monotonically. (B) Representative tip trajectories generated using the forward kinematics formulation for segmented mechanisms in Eq. 5 . Trajectories are for $J=3$ sections with distinct aspect ratios ( $\alpha_{j}=\{0.41,0.17,0.55\}$ for solid line, $\alpha_{j}=\{0.44,0.43,0.23\}$ for short dashed line, $\alpha_{j}=\{0.71,0.32,0.51\}$ for long dashed line, $\alpha_{j}=\{0.41,0.62,0.89\}$ for dot-dashed line). The mechanism is capable of tracing complex paths with varying curvature, including spirals and self-intersecting loops. (C) The mechanism is parameterized by constant aspect ratios $\{\alpha_{j}\}$ (function in green), which determine the tip trajectory $\mathbf{r}_{\textrm{tip}}(\Psi)$ (shown in red) through the forward kinematic relations in Eq. 5 . The resulting trajectory is compared against a prescribed target curve (blue) in the tip-trajectory error $\mathcal{L}_{\textrm{tip}}$ (see Eq. 6 ), which quantifies the curvature mismatch between $\kappa_{\textrm{tip}}(s;\{\alpha_{j},N_{j}\},\Psi)$ and the target curvature $\kappa^{t}(s)$ . The differentiable simulation provides exact derivatives (such as $\partial\mathbf{r}_{\textrm{tip}}/\partial\alpha_{j}$ ) via automatic differentiation, enabling efficient gradient-based updates of $\{\alpha_{j}\}$ (for fixed $\{N_{j}\}$ ) and the actuation input $\Psi$ range. (D) Snapshots of the scissor mechanism’s deployment as the tip traces a circular trajectory. The red colored stroke indicates the tip trajectory, and the inset shows the aspect-ratio, $\{\alpha_{j}\}$ obtained using the optimization procedure. (E) Comparison between target curves (blue) and tip trajectories (red) when the mechanism traces the characters $j$ and $D$ , as a tribute to the automaton artist Jaques-Droz. The color gradient from light to dark indicates tracing direction as $\Psi$ decreases from $\Psi_{\max}$ to $\Psi_{\min}$ . Insets show the optimized aspect-ratio profiles $\{\alpha_{j}\}$ , where $j$ indexes the scissor-units. See SI Video 3 for a visualization of the mechanism performing the writing task.

Figure S1 : (A) Three scissor-units with the same aspect ratio are assembled by connecting neighboring units. Their center nodes are marked by the yellow points $\mathbf{r}_{j-1}$ , $\mathbf{r}_{j}$ , and $\mathbf{r}_{j+1}$ . Using the turning-angle definition, the curvature is given by the angle $\vartheta_{j}$ between the vectors joining adjacent center nodes, namely $\mathbf{v}_{j-1}$ and $\mathbf{v}_{j}$ (visualized in green). The associated length scale is defined as $|\mathbf{v}_{j-1}|+|\mathbf{v}_{j}|$ . (B) For a single scissor-unit, the green circle is the osculating circle of radius $R_{\mathrm{osc}}$ ; the member orientation vectors $\hat{\mathbf{t}}^{(1)}$ and $\hat{\mathbf{t}}^{(2)}$ are tangent to this circle. (C) A two-segment scissor mechanism for an arbitrary choice of the base center node position $\mathbf{r}_{0}$ and the initial orientation $\hat{\mathbf{p}}_{1}$ . For a given actuation angle $\Psi$ , each section $j$ is modeled as a circular arc with center of curvature $\tilde{\mathbf{q}}_{j}$ and radius $\kappa_{j}^{-1}$ . At the interface between sections (node $\mathbf{r}_{1}$ ), the change in aspect ratio from $\alpha_{1}$ to $\alpha_{2}$ requires a shift $\zeta_{j}^{j+1}$ between successive centers of curvature to maintain geometric continuity (from $\tilde{\mathbf{q}}_{1}$ to $\tilde{\mathbf{q}}_{2}$ ). The final tip position $\mathbf{r}_{\text{tip}}$ is obtained by iterating these section-wise rotations and translations.

Figure S2 : (A) Scissor mechanism corresponding to $N=30$ with $\alpha=0.52$ and $\Psi=\pi/4$ . (B) Comparison between shape of scissor mechanism with $N=30$ and $\Psi=\pi/4$ constructed from full non-linear simulation (shown in blue) for $\alpha_{j}=\alpha_{0}+\epsilon j$ and perturbative solution in Eq. ( S31 ) (shown in gray). We see a good match between the perturbative solution and the full solution. We have chosen $\alpha_{0}=0.52,\epsilon=0.001$ .

Figure S3 : Algorithm used for the differentiable framework for the writing task explained in Sec. S6 . The four steps involved in this framework are:

(i)

Target processing - normalization of raw target coordinates to a unit bounding box followed by arc-length parameterization to extract the smoothed target curvature profile

\kappa^{t}(s)

;

(ii)

Differentiable forward model - evaluation of tip trajectory via forward kinematics and subsequent reparameterization from

\Psi

to arc length

s

to obtain the intrinsic mechanism curvature

\kappa_{\text{tip}}(s^{M})

as a differentiable function of arc length;

(iii)

Gradient descent - iterative optimization of design parameters

\bm{\theta}

using the Adam optimizer, with exact gradients of the composite loss

\mathcal{L}[\bm{\theta}]

computed via automatic differentiation;

(iv)

Arc-length alignment - differentiable linear interpolation of the mechanism curvature onto the fixed target arc-length grid to enables a pointwise error comparison.

Figure S4 : Collage of the tip trajectories for a mechanism with $N=50$ and 15 independent realizations using random initializations. The minimum (Run 13, Loss: 0.5329) is selected as the optimal configuration, while other runs capture varying degrees of target character fidelity or converge to suboptimal local minima.

Figure S5 : Collage of the tip trajectories from 15 independent optimization runs for a fixed number of units $N=8$ . While many runs converge to the target circle with low loss (e.g., Run 6, Loss: 0.0582), some realizations fall into local minima with high loss (e.g., Run 8, Loss: 5.8510), highlighting the complex optimization landscape.

Figure S6 : (A) The plot displays the sensitivity

\sigma_{j}

, defined as the variance of the tip displacement distribution

p(\Delta r^{(j)})

, against the normalized unit index

j/N

. Results for different mechanism sizes (

N=100,200,300

) collapse onto a single curve, showing that the sensitivity profile is independent of the total number of units when scaled by the mechanism’s length. The data indicates that the distal tip position is most sensitive to perturbations in units located near the base (

j/N\to 0

) and becomes progressively more robust to geometric errors as the unit index approaches the tip (

j/N\to 1

Figure S7 : Three-dimensional view of CAD model of the controller-actuator setup. This setup includes a high-torque servo that drives the actuating unit; an Arduino Nano microcontroller processing logic; a potentiometer for manual angle control; a start/stop button for simple triggering; and an AC connector for stable external power.

Mechanism Design

Mechanism design studies rules for allocating resources and shaping incentives when participants hold private information or act strategically.

See all 142 papers

Oct 11, 2026

Kinematic Reward, Measured Effort: Hardware-Validated RL Morphology Optimization of a Planar 2R Manipulator

Yoshikawa's Manipulability Index is generally used as a classical measure to define an optimal morphology of a manipulator. However, any such optimization is almost always studied as a purely kinematic quantity in simulation and is rarely validated on hardware. We present a reinforcement-learning framework with RSL-RL PPO that optimizes the morphology of a 2R SCARA-type manipulator arm with Yoshikawa's kinematic manipulability index as a core reward, and the analytically derived closed-form optimum for the circular task is used to verify that our method recovers it. Then the same method is applied to custom elliptical and rectangular paths having no analytical optimum. The RL-optimized morphologies are then transferred to a 2R hardware manipulator to validate its effect in reduction of joint load in the hardware. Compared to the worst morphology, the optimum morphology delivered by our method reduces second joint load by 38.4 +/- 0.6% for the ellipse path and 41.3 +/- 0.3% for the rectangle, and first joint load by 15%.

Gunesh Gupta, Arvind Kumar Mishra, Sohom Chakrabarty

Independent Researcher, Haridwar, India · Department of Electrical Engineering, Indian Institute of Technology Roorkee, Roorkee, India

Robotic Optimization Robotic Manipulation Robotic RL

Ask PDF

Apr 6, 2026

ZipFold: Modular Actuators for Scaleable Adaptive Robots

There is a growing need for robots that can change their shape, size and mechanical properties to adapt to evolving tasks and environments. However, current shape-changing systems generally utilize bespoke, system-specific mechanisms that can be difficult to scale, reconfigure or translate from one application to another. This paper introduces a compact, easy-to-fabricate deployable actuator that achieves reversible scale and stiffness transformations through compound folding and zipping of flexible 3D-printed plastic strips into square-section deployable beams. The simple actuation method allows for smooth, continuous transitions between compact (flexible) and expanded (quasi-rigid) states, facilitating diverse shape and stiffness transformations when modules are combined into larger assemblies. The actuator's mechanical performance is characterized and an integrated system involving a four-module adaptive walking robot is demonstrated.

Niklas Hagemann, Daniela Rus

Computer Science and Artificial Intelligence Laboratory, MIT, Cambridge, MA 02139 USA

Modular Robotics

Ask PDF

Sep 9, 2026

Assembling Two Parts in One Hand

A hallmark of human dexterity is the cooperative use of fingers, where different fingers take on distinct yet coordinated roles to accomplish fine manipu- lation, such as capping a pen with the hand that holds it. We study this finger-level coordination through in-hand assembly: mating two rigid objects within a single dexterous hand, with no second arm and no fixture. We present a reinforcement learning formulation to solve this problem in a unified framework, which is driven by a goal relative pose between the two parts. Finger coordination is shaped by a function-based auxiliary reward and regularized toward a single human reference pose, while domain randomization and a fusion of historical proprioception and object observation confer robustness to occlusion-induced estimation noise. The same recipe solves three different assembly tasks (Bottle, Syringe, and Marker). Trained purely in simulation, the policies transfer zero-shot to hardware with a single camera, demonstrating robustness to state-estimation errors caused by oc- clusion. Our experiments also reveal that in-hand assembly places demands on hand morphology and can serve as a benchmark for modern robotic hand systems. Videos and code are available at https://ltbgbird.github.io/in-hand-assembly-page/.

Liuao Pei, Tianyue Wu, Hui Zhang +2

The University of Hong Kong · The Hong Kong University of Science and Technology (Guangzhou) · ETH Zurich +1

RL for Robotics Dexterous Manipulation In-Hand Manipulation Robotic Assembly Hand-Object Interaction

Ask PDF

Abstract

Mechanism Design

Explore similar work

Kinematic Reward, Measured Effort: Hardware-Validated RL Morphology Optimization of a Planar 2R Manipulator

ZipFold: Modular Actuators for Scaleable Adaptive Robots

Assembling Two Parts in One Hand