Series-Elastic Delta Parallel Manipulator

• The paper presents an MSA-based stiffness modeling approach that integrates elastic joint and flexible link representations for Delta-type architectures.
• It details how kinematic closure constraints and Lagrange multipliers are applied to derive the complete Cartesian stiffness matrix for precise compliance analysis.
• The method provides a systematic framework for evaluating actuator series elasticity, supporting advanced design and sensitivity studies in parallel robotics.

A series-elastic Delta-type parallel manipulator is a class of closed-loop kinematic robotic architectures wherein the primary actuators at the base joints include intrinsic series elasticity. This system combines the geometric and structural characteristics of the classical Delta manipulator—three identical kinematic chains connecting fixed and moving triangular platforms—with revolute actuators exhibiting non-negligible compliance about their axes. The Matrix Structural Analysis (MSA) framework provides an analytical methodology for deriving the complete stiffness model of such manipulators, accommodating both rigid and flexible links, multiple forms of joints (rigid, passive, or elastic), and explicit kinematic closure constraints dictated by the parallel architecture (Klimchik et al., 2018).

1. Link and Joint Element Modeling in MSA

In the MSA approach, each robotic limb is decomposed into element-level stiffness representations. Flexible links, denoted as $\ell$ with nodes $i, j$, are modeled by a $12 \times 12$ stiffness matrix $K_\ell$ relating nodal deflections $\Delta t_i,\ \Delta t_j \in \mathbb{R}^6$ to end wrenches $W_i,\ W_j \in \mathbb{R}^6$:

$$\begin{bmatrix} W_i \ W_j \end{bmatrix} = K_\ell \begin{bmatrix} \Delta t_i \ \Delta t_j \end{bmatrix},$$

with $K_\ell$ partitioned into $6 \times 6$ blocks. For ideally rigid links, the connection is enforced by compatibility constraints:

$\Delta t_j = T_{ij} \Delta t_i,$

and force equilibrium:

$W_i + T_{ij}^T W_j = 0,$

equivalent to the infinite stiffness limit for $K_\ell$ in those degrees corresponding to $T_{ij}$.

A series-elastic revolute actuator, regarded as an “elastic joint,” connects two nodes ($i$, $j$) and is parameterized by an axis selector $A \in \mathbb{R}^{1 \times 6}$ and joint stiffness $K_a$. The axial deflection is $\Delta d := A(\Delta t_j - \Delta t_i)$ and the developed wrench $W_e := K_a \Delta d$. Represented in nodal form:

$$\begin{bmatrix} W_i \ W_j \end{bmatrix} = K_j \begin{bmatrix} \Delta t_i \ \Delta t_j \end{bmatrix}, \quad K_j = \begin{bmatrix} A^T K_a A & -A^T K_a A \ -A^T K_a A & A^T K_a A \end{bmatrix} \in \mathbb{R}^{12 \times 12}.$$

2. Kinematic Closure Conditions for Delta-Type Architecture

The Delta manipulator consists of three identical limbs, each terminating at a node affixed to the common moving platform. Denoting the translational deflection of the $i$-th limb’s endpoint as $\Delta t_{P,i} \in \mathbb{R}^3$ and rotation as $\Delta \varphi_{P,i} \in \mathbb{R}^3$, all must be congruent to the single platform’s global deflection $$\Delta x_P = [\Delta t_P;\ \Delta \varphi_P] \in \mathbb{R}^6$$:

$$\Delta t_{P,i} - \Delta t_P = 0, \quad \Delta \varphi_{P,i} - \Delta \varphi_P = 0, \quad i = 1,2,3.$$

These yield $18$ scalar constraints, expressible in block-sparse matrix form as $C u = 0$, where $u$ is the global vector of nodal deflections.

$$C = \begin{pmatrix} I_6 & 0 & 0 & -I_6 \ 0 & I_6 & 0 & -I_6 \ 0 & 0 & I_6 & -I_6 \end{pmatrix}, \quad C u = 0.$$

3. Global System Assembly: Equilibrium and Compatibility

To derive the manipulator’s global response, all element stiffness contributions are assembled using Boolean matrices ($T_\ell$, $T_j$) embedding each element's $12$-DOF local behavior into the overall nodal ordering:

$$K_{ee} = \sum_{\ell \in E} T_\ell^T K_\ell T_\ell + \sum_{j \in J} T_j^T K_j T_j, \quad K_{ee} \in \mathbb{R}^{n_u \times n_u}$$

where $E$ indexes flexible elements and $J$ all elastic joints, including actuators.

Constraints are imposed via Lagrange multipliers $\lambda \in \mathbb{R}^{n_c}$, assembling the full system: $$\begin{pmatrix} K_{ee} & C^T \ C & 0 \ \end{pmatrix} \begin{pmatrix} u \ \lambda \end{pmatrix} = \begin{pmatrix} f \ 0 \end{pmatrix},$$ where $f$ collects applied nodal wrenches.

4. Modeling Series Elasticity in the System Stiffness

Series elasticity at the actuators is introduced by incorporating respective joint blocks $K_{a,i}$ in the assembly. For three limb base joints,

$$K_{ee} = K_{ee}^0 + \sum_{i=1}^3 T_{a,i}^T K_{a,i} T_{a,i},$$

where $K_{ee}^0$ is the system stiffness in the absence of actuator compliance, and each $K_{a,i}$ is the $12 \times 12$ joint stiffness matrix for actuator $i$. This construction ensures that the only alteration to the manipulator's stiffness model, relative to the rigid-actuator case, is the addition of these specific compliance contributions.

5. Solution Strategy and Cartesian Stiffness Extraction

Solving the constrained linear system, the vector of internal Lagrange multipliers $\lambda$ is eliminated via block-Gaussian elimination: $$u = K_{ee}^{-1}(f - C^T \lambda), \quad \lambda = (C K_{ee}^{-1} C^T)^{-1} C K_{ee}^{-1} f.$$ This yields $u(f)$, the complete distribution of nodal deflections under load.

To relate the platform's six-dimensional wrench $f_P$ to its deflection $\Delta x_P$, a $6 \times n_u$ selection matrix $J$ isolates these DOFs:

$$\Delta x_P = J u = J K_{ee}^{-1} (I - C^T (C K_{ee}^{-1} C^T)^{-1} C K_{ee}^{-1}) f = K_{\text{cart}}^{-1} f_P.$$

The Cartesian stiffness matrix, mapping end-effector wrench to deflection, is

$$K_{\rm cart} = \left(J K_{ee}^{-1} J^T\right)^{-1},$$

where all link and series-elastic actuator stiffnesses are included in $K_{ee}$ and $J$ restricts to the platform node’s DOFs.

6. Mathematical Summary and Relevance

The MSA-based approach to stiffness modeling—a direct aggregation of link and joint stiffness blocks, merged with kinematic closure and constraint equations—affords systematic and extensible analysis of complex parallel manipulator architectures with series compliance. Notably, it avoids conventional matrix row/column merging but instead applies constraints and equilibriums through global sparse matrix inversion, streamlining the procedure for manipulators with mixed joint and link characteristics (Klimchik et al., 2018).

Equation cross-references within the cited methodology:

• Flexible link element: Eq. (1)
• Elastic joint model: Eq. (10)
• Global system assembly: Eq. (18–19)
• Lagrange multiplier elimination and Cartesian stiffness: Eq. (21)

---

The described methodology provides a rigorous, implementable stiffness modeling framework for Delta-type parallel manipulators with revolute series-elastic actuators, making it suitable for advanced structural analysis and sensitivity studies in robotics research and compliant mechanism design (Klimchik et al., 2018).