Quasi-Static Shape Control

• Quasi-static shape control is the regulation of geometry through slowly varying, equilibrium-constrained manipulations that neglect inertial effects.
• It has been applied across nonlinear structures, soft robots, and free-boundary fluids using tools like finite element analysis, feedback linearization, and experimental continuation.
• Key insights include equilibrium discovery, deformation-path optimization, and the challenges of multi-stability, non-local coupling, and instability during state transitions.

Searching arXiv for recent and foundational papers on quasi-static shape control and closely related formulations. Quasi-static shape control denotes the regulation, planning, or traversal of shape under slowly varying actions or requirements, with inertia neglected or treated as secondary and the controlled state interpreted through equilibrium constraints. Across the literature, the term covers nonlinear structures loaded through a main actuation point and auxiliary probes, compliant and soft bodies controlled in strain or task coordinates, manipulation systems described by friction-dominated kinematics or contact complementarity, and free-boundary fluid models in which the “shape” is a time-evolving domain or contact line (Neville et al., 2018, Fehér et al., 16 Sep 2025, Siddharth, 2 May 2026, Chau et al., 2022). A common thread is that control acts by selecting equilibria, stabilizing equilibrium manifolds, or navigating stable sub-manifolds of the equilibrium set rather than exploiting inertial transients (Fehér et al., 16 Sep 2025).

1. Modeling regime and scope

In the strictest sense, quasi-static modeling assumes that the system evolves slowly enough that inertial effects are negligible relative to elastic, gravitational, frictional, capillary, or contact forces. In planar slider-pusher manipulation, this is stated as the regime in which inertial forces are negligible compared to frictional forces, so the system admits a force-motion relationship instead of a full dynamic model (Witte et al., 6 Nov 2025). In sessile-drop relaxation, the governing assumption is that the energy dissipated near the contact line is much larger than that in the bulk of the fluid, so the evolution can be treated as a dissipative mechanical system with quasi-static surface equilibration (Iliev et al., 2016). In motion design for adaptive compliant structures, the quasi-static setting is explicitly described as involving no inertia or kinetic terms (Sachse et al., 2020).

These formulations do not imply a single state representation. Depending on the application, the controlled object may be a displacement field, a finite-dimensional strain vector, a rod or shell embedding, a contact-rich object pose, a path-following manifold, or a fluid domain.

Area	Controlled entity	Representative formulation
Nonlinear structures	Overall shape under main actuation and probes	Equilibrium identified by $P_\mathrm{p}=0$ (Neville et al., 2018)
Soft rods and shells	Strain, tip pose, or shell embedding	Quasi-static equilibrium under wrench or magnetic actuation (Siddharth, 2 May 2026, Javadi et al., 4 Oct 2025)
Manipulation	Object pose under frictional contact	Limit-surface or LCP-based quasi-static map (Witte et al., 6 Nov 2025, Halm et al., 2019)
Free boundaries	Domain $\Omega_t$ or contact line	Boundary motion given by a local velocity law (Chau et al., 2022, Iliev et al., 2016)

Taken together, these works suggest that “shape control” in the quasi-static regime is best understood as equilibrium-constrained geometry control rather than as a single algorithmic family.

2. Equilibrium sets, stability, and geometric structure

A general formulation appears in "Quasi-static shape control of soft, morphing structures" (Fehér et al., 16 Sep 2025). There, the equilibrium set is

$\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$

with dimension

$d = d_a + d_l.$

The stable sub-manifold consists of equilibria satisfying the second-variation condition

$D^2L(\tilde{u}) > 0.$

Because finite deformations are allowed, the equilibrium set may exhibit bifurcations, self-intersections, barriers, gaps, and multiple connected components; quasi-static shape adaptation is therefore described as navigation along the stable sub-manifold under slow changes of actuator parameters and loads (Fehér et al., 16 Sep 2025).

This geometric viewpoint is mirrored in "A Basic Mechanical and Geometric Framework for Quasi-Static Manipulation" (Campolo et al., 2023), which places quasi-static manipulation in force-space, the cotangent bundle $T^*Q$ of the configuration space, and interprets equilibrium pairs as Lagrangian submanifolds. For a conservative potential $W$, the control Hessian

$$G_m(u) = \nabla^2_{uu} W - \nabla^2_{zu} W (\nabla^2_{zz} W)^{-1} \nabla^2_{uz} W$$

is generally symmetric but not necessarily positive definite, so the paper proposes the squared-Hessian metric

$G_m^2(u) = G_m(u)^T G_m(u),$

together with the path functional

$$J = \int_0^1 \sqrt{ \dot{u}^T G_m^2(u)\, \dot{u} }\, ds.$$

This suggests a unifying interpretation of quasi-static shape control as path planning on an equilibrium manifold endowed with a mechanically derived metric (Campolo et al., 2023).

The same equilibrium-set perspective immediately clarifies a recurring misconception: quasi-static evolution does not imply uniqueness of the controlled shape. Multi-stability, disconnected stable branches, and singular Jacobians are structural features rather than modeling defects (Fehér et al., 16 Sep 2025).

3. Structural mechanics and experimental continuation

In nonlinear structural mechanics, quasi-static shape control emerged as a means to access equilibria that are invisible under conventional single-point loading. "Shape Control for Experimental Continuation" (Neville et al., 2018) studies a shallow arch loaded at a midpoint while its overall shape is controlled by two additional probes. The central observation is that, for a given midpoint displacement $\Omega_t$0, the midpoint reaction force $\Omega_t$1 depends on the overall shape of the structure. Probe scans are performed by fixing $\Omega_t$2, actuating the probes, and monitoring the probe reaction $\Omega_t$3; whenever $\Omega_t$4, the structure is in equilibrium for that global shape. Repeating these scans at different midpoint displacements makes unstable equilibria experimentally accessible, and unstable segments of the equilibrium path were identified experimentally for the first time (Neville et al., 2018).

This framework turns shape into the experimental analogue of the additional continuation parameter that numerical arc-length methods provide. It also shows that quasi-static shape control is not limited to stabilization in the neighborhood of a stable configuration; it can be used to reveal unstable branches of the equilibrium manifold itself (Neville et al., 2018).

A complementary formulation appears in "A variational formulation for motion design of adaptive compliant structures" (Sachse et al., 2020). There, the objective is not merely to find isolated equilibria, but to design quasi-static motions between prescribed geometrical configurations by minimizing a path functional, with the exemplar choice

$\Omega_t$5

The method uses two finite element discretizations: the standard spatial discretization and an additional discretization of the deformation path. The resulting nonlinear system is solved monolithically by Newton-Raphson, and analytical sensitivity analysis is available from standard nonlinear finite element components (Sachse et al., 2020). Benchmark examples include rigid body motions, instability phenomena, and inextensible deformations of shells. A plausible implication is that quasi-static shape control in structures naturally splits into two tasks: equilibrium discovery and deformation-path optimization.

4. Soft robots, rods, shells, and deformable bodies

For soft robots modeled as Cosserat bodies, quasi-static shape control is often formulated directly in strain or task space. "Quasi-Static Control of Discrete Cosserat Rod" (Siddharth, 2 May 2026) uses the Piecewise Constant Strain discretization, so the rod state is

$\Omega_t$6

with quasi-static statics equation

$\Omega_t$7

State-feedback linearization is developed in both strain and task spaces, with the external end-effector wrench as control input, and global asymptotic stability is established for the proposed feedback laws (Siddharth, 2 May 2026). In this setting, shape regulation and end-effector trajectory tracking are treated within one static model.

A shell analogue appears in "Geometrically Exact Hard Magneto-Elastic Cosserat Shells: Static Formulation for Shape Morphing" (Javadi et al., 4 Oct 2025). The shell configuration manifold is the space of smooth embeddings $\Omega_t$8, and the local deformation measure is defined through the Lie-group quantity

$\Omega_t$9

The strong and weak forms of equilibrium are derived from the principle of virtual work, and the finite element implementation is designed to avoid singularity and locking phenomenon in modeling shell structures (Javadi et al., 4 Oct 2025). The control significance is explicit: regulating the external magnetic field yields direct actuation over the shell’s quasi-static shape.

Manipulation models of friction-dominated bodies provide another branch of quasi-static shape control. "Differential Flatness of Quasi-Static Slider-Pusher Models with Applications in Control" (Witte et al., 6 Nov 2025) derives a differential kinematic model from the limit surface approach under the quasi-static assumption and with negligible contact friction. For polygon sliders and circular pushers, the centre of mass is a flat output, enabling both a cascaded quasi-static feedback strategy and a dynamic feedback linearization approach (Witte et al., 6 Nov 2025). For cloth manipulation, "QDP: Learning to Sequentially Optimise Quasi-Static and Dynamic Manipulation Primitives for Robotic Cloth Manipulation" (Blanco-Mulero et al., 2023) treats quasi-static primitives such as pick-and-place and drag as parameterized actions whose height, move time, or velocity materially affect the resulting cloth shape; sequential reinforcement learning is used to select these parameters, and simulation performance improves by 20% compared to sub-optimal choices (Blanco-Mulero et al., 2023).

These examples show that, in robotics, quasi-static shape control ranges from model-based equilibrium regulation to data-driven parameter selection, but remains anchored in slow, geometry-dominated evolution.

5. Quasi-static feedback, flatness, and path-following

In nonlinear control, “quasi-static feedback” often denotes a feedback law that uses the classical state and a finite number of derivatives of a new input, without increasing system order. "Tracking Control for $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$0-Flat Systems by Quasi-Static Feedback of Classical States" (Gstöttner et al., 2021) constructs such feedback for $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$1-flat systems. The key input-output relation is

$\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$2

achieved by a feedback of the form

$\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$3

Tracking is then obtained by choosing $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$4 so that each flat-output component satisfies a linear asymptotically stable error equation (Gstöttner et al., 2021). The contribution is practical as well as structural: generalized Brunovský states are avoided, and feedback depends only on classical measurable states.

"Exact Linearization of Minimally Underactuated Configuration Flat Lagrangian Control Systems by Quasi-Static Feedback of Classical States" (Hartl et al., 2023) specializes this logic to configuration-flat Lagrangian systems with $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$5 degrees of freedom and $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$6 inputs. The closed-loop system is transformed into decoupled integrator chains

$\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$7

and the paper identifies the multi-indices that permit rest-to-rest transitions: one component of $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$8 must be $\mathbb{E} = \left\{ (u, a, l) \ \Big|\ \text{structure in equilibrium at configuration %%%%0%%%% under actuator parameters %%%%1%%%% and load parameters %%%%2%%%%} \right\},$9, and the remaining components must be $d = d_a + d_l.$0 (Hartl et al., 2023). This is directly relevant to shape control because quasi-static maneuvers are commonly rest-to-rest transitions between static configurations.

A related but application-specific interpretation appears in "Path-following Control of a Quadrotor using Quasi-Static Transverse Feedback Linearization" (Lawati et al., 22 Jun 2026). Here the controlled object is a geometric path rather than a deformable body. The path-following manifold

$d = d_a + d_l.$1

is rendered invariant, thrust is computed algebraically from the current state, and only a $d = d_a + d_l.$2 decoupling matrix must be inverted for the torque inputs (Lawati et al., 22 Jun 2026). The result is local exponential stability of the path-following manifold. This broader control usage indicates that “shape control” can also mean regulation of geometric path shape rather than only material deformation.

6. Free-boundary liquids, non-locality, and the limits of local control

Free-boundary fluid models expose two sharp limits of quasi-static intuition: convexity need not be preserved, and local boundary laws do not imply local controllability. In "Instantaneous convexity breaking for the quasi-static droplet model" (Chau et al., 2022), the droplet occupies a time-evolving domain $d = d_a + d_l.$3, with height $d = d_a + d_l.$4 governed by

$d = d_a + d_l.$5

The paper constructs a smooth convex initial domain in $d = d_a + d_l.$6 that instantly loses convexity. The mechanism is local curvature of the boundary velocity $d = d_a + d_l.$7, together with the condition

$d = d_a + d_l.$8

which makes $d = d_a + d_l.$9 strictly positive near edge endpoints in the construction (Chau et al., 2022). Convexity is therefore fragile, not invariant.

"Quasi-static relaxation of arbitrarily shaped sessile drops" (Iliev et al., 2016) derives the local contact-line law

$D^2L(\tilde{u}) > 0.$0

from a dissipation functional proportional to the contact line length. Yet the same work shows that relaxation at a given point of the contact line is influenced by the dynamics of the whole drop, which is described as a manifestation of the non-local character of the problem (Iliev et al., 2016). A common misconception is therefore ruled out twice: quasi-static evolution may be specified by local constitutive laws, but the resulting shape dynamics can remain globally coupled and can even destroy geometric properties such as convexity.

The validity of the quasi-static assumption itself is also application-dependent. "Model-based Manipulation of Deformable Objects with Non-negligible Dynamics as Shape Regulation" (Tiburzio et al., 2024) explicitly contrasts traditional quasi-static methods, which apply to objects that are extremely lightweight, small, or constrained to move very slowly, with a fully dynamic framework for slender deformable objects grasped at one end. In experiments on six different electric cables, the model-based dynamic approach expanded the reachable workspace and reduced average endpoint position error by 58–78% compared to a quasi-static reference method (Tiburzio et al., 2024). This does not negate quasi-static shape control; it identifies its operational boundary.

Across these domains, quasi-static shape control is most accurately characterized as equilibrium-governed geometry control under slow variation. Its principal strengths are structural clarity, computational tractability, and access to stability and path-planning questions on equilibrium manifolds. Its principal limitations are equally structural: multi-stability, singularities, unstable branches, non-local coupling, and breakdown when inertia or fast transients become essential.