cs.ROMar 17, 2026

Reconciling distributed compliance with high-performance control in continuum soft robotics

Authors: Vito Daniele Perfetta, Daniel Feliu-Talegon, Ebrahim Shahabi, Cosimo Della Santina

Organizations: Cognitive Robotics Department, Faculty of Mechanical Engineering, Delft University of Technology, Delft 2628 CD, The Netherlands.

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

High-performance closed-loop control of truly soft continuum manipulators has remained elusive. Experimental demonstrations have largely relied on sufficiently stiff, piecewise architectures in which each actuated segment behaves as a distributed yet effectively rigid element, while deformation modes beyond simple bending are suppressed. This strategy simplifies modeling and control, but sidesteps the intrinsic complexity of a fully compliant body and makes the system behave as a serial kinematic chain, much like a conventional articulated robot. An implicit conclusion has consequently emerged within the community: distributed softness and dynamic precision are incompatible. Here we show this trade-off is not fundamental. We present a highly compliant, fully continuum robotic arm - without hardware discretization or stiffness-based mode suppression - that achieves fast, precise task-space convergence under dynamic conditions. The platform integrates direct-drive actuation, a tendon routing scheme enabling coupled bending and twisting, and a structured nonlinear control architecture grounded in reduced-order strain modeling of underactuated systems. Modeling, actuation, and control are co-designed to preserve essential mechanical complexity while enabling high-bandwidth loop closure. Experiments demonstrate accurate, repeatable execution of dynamic Cartesian tasks, including fast positioning and interaction. The proposed system achieves the fastest reported task-execution speed among soft robots. At millimetric precision, execution speed increases nearly fourfold compared with prior approaches, while operating on a fully compliant continuum body. These results show that distributed compliance and high-performance dynamic control can coexist, opening a path toward truly soft manipulators approaching the operational capabilities of rigid robots without sacrificing morphological richness.

Figures & tables

  SegmentedNon-segmented BendingGeneral
  [ 13 ][ 10 ][ 14 ][ 9 ][ 11 ][ 20 ][ 21 ]This Work
Robot↑\uparrow Material softness■■□□\blacksquare\blacksquare\square\square■■□□\blacksquare\blacksquare\square\square■■□□\blacksquare\blacksquare\square\square■■■□\blacksquare\blacksquare\blacksquare\square■■■□\blacksquare\blacksquare\blacksquare\square■□□□\blacksquare\square\square\square■■□□\blacksquare\blacksquare\square\square■■■■\blacksquare\blacksquare\blacksquare\blacksquare
Deformation strains        
Torsion×\times×\times×\times×\times×\times×\times×\times✓
Bending✓✓✓✓✓✓✓✓
Axial✓✓✓✓×\times×\times×\times✓
Table 1: Comparison of the proposed work with the most relevant experimental results on tas k-space dynamic control of soft robots. Results are organized for segmented, non-segmented reduced-compliance bending, and general continuum soft structures, reporting deformation strains, material softness, model and control techniques † , generalizability, preliminary training requirements, stability guarantee (with ∼\sim indicating limited validity), task-level feedback, 3D task achievement, and the Root Mean Square of the Task-Space error. Finally, task velocity is given for millimeter-level (errors <1<1 cm) and centimeter-level (errors <3<3 cm) precision scenarios, where dashed lines indicate unmet performance criteria and (⋅)∗(\cdot)^{*} denotes trajectory reference velocity when point-to-point regulation is absent and actual speed values are not reported. Additional details are reported in the Supplementary Material.
Figure 1: From Cobots to Soft Robots. Comparison of physical characteristics across collaborative rigid robots, segmented soft robots, and general soft robots. The table highlights key differences in structure, material softness, actuation, and motion capabilities.
Figure 2: Experimental Setup and Control Framework. ( A ) Experimental setup. ( B ) Internal view of the soft arm. Different tendon configurations are visually distinguished by straight (green) and crossed (blue) paths, while burgundy circles denote the tiny guiding rings. Transparent renderings on the sides illustrate examples of achievable shapes. ( C ) Top view of the robot illustrating the tendons’ placement at the base. Each tendon–motor pair is identified by a number from 1 to 4, with green circles denoting straight tendon paths and blue crosses denoting the crossed ones. ( D ) Soft robot’s workspace attained through combined actuation of the four tendons with pulling forces in [0,−5]\left[0,-5\right] N. ( E ) Block diagram of the two-level control framework employed, with the inner loop in gray and the outer loop in red.
Figure 3: Two-dimensional coiling task. ( A ) (Top) The robot’s final configurations at each reference Cartesian coordinate pair are shown. Reference points are marked with red circles, while the corresponding 3D projections reached by the robot are highlighted with red stars. From right to left, the shapes correspond to time windows of 1515 s, 55 s and 11 s, respectively, with a transparency gradient representing progression over time. (Bottom) The projection of the robot’s motion onto the Cartesian plane is shown for each task and time interval. Red circles denote desired target coordinates, while black crosses indicate the robot’s final position before reference change. Motion is reported on a normalized time scale. ( B ) Time evolution of the errors in operational space and their convergence within 55 mm and 1010 mm bands. To improve readability, the post-processed error obtained through a moving-average filter is highlighted, while the raw error signal is displayed more transparently in the background. Grey dashed vertical bars indicate the switch in time windows. ( C ) Time evolution of the estimated actuation coordinates (blue) and their tracking of the reference signals (red) under configuration-space collocated control.
Figure 4: Four-dimensional triangular shape task. ( A ) (Top) The robot’s final configurations upon converging at reference points (red circles) displaced in a triangular shape in the Cartesian space are shown. From right to left, the shapes correspond to time windows of 2525 s, 55 s and 11 s, respectively. (Bottom) The reconstructed robot’s motion in the Cartesian space is shown for each time interval. The red circle denotes the desired targets, while the black cross indicates the robot’s last position before the reference change. Motion is reported on a normalized time scale. ( B ) Time evolution of the errors in operational space and their convergence within the 55 mm and 1010 mm accuracy bands. To improve readability, the post-processed error obtained through a moving-average filter is highlighted, while the raw error signal is displayed more transparently in the background. Grey dashed vertical bars indicate the switch in time windows. ( C ) Time evolution of the pulling forces exerted by the tendons. ( D ) Time evolution of the estimated actuation coordinates (blue) and their tracking of the reference signals (red) under configuration-space collocated control.
Figure 5: Pendulum-like ball striking. ( A ) The time sequence of shapes attained by the soft manipulator is reported here. The ball is first released at t=0.00t=0.00 s, and then accurately struck at t=0.35t=0.35 s by the robot. ( B ) The motion of the robot tip together with the trajectory of the marker placed on the external surface of the pendulum is shown. The corresponding normalized time histories are reported on the left (soft robot) and on the right (pendulum). The trajectories include cross markers, shown in the corresponding colors, to identify the impact instant; post-impact trajectories are subsequently displayed with increased transparency. In addition, the desired target position for the robot’s tip is indicated by the gray circle, while the nominal trajectory of the pendulum’s marker in the absence of impact is depicted by the black dashed line.
Figure 6: Sugar pick-and-pour task to sweeten a cup of coffee. ( A ) The sequence of shapes attained by the soft manipulator is here reported for t=42,60,98,102t=42,60,98,102 s. Specifically, the arm accesses the sugar container at t=42t=42 s, and eventually releases it at t=102t=102 s. The starting configuration of the robot is shown transparently in the top-left picture. ( B ) The time evolution of the errors in operational space is shown, together with their convergence within the 1010 mm and 3030 mm accuracy bands. To improve readability, the post-processed error obtained through a moving-average filter is highlighted, while the raw error signal is displayed more transparently in the background. ( C ) The time evolution of the estimated actuation coordinates (blue) and their tracking of the reference signals (red) under configuration-space collocated control is shown.
More in this topic

Continuum Robot Control

Continuum robot control studies how to model, sense, and guide flexible robots whose bodies bend and deform along their length.

See all 70 papers

Explore similar work

CardsList
Jun 29, 2026

Real-Time Compliance and Position Control of a Hyper-redundant Soft Robotic Arm

Robots working in unstructured or partially unobservable environments must combine accurate motion with physical compliance that can passively correct contact misalignment. Soft robots provide this compliance but have struggled to precisely control their tip compliance and position. This paper presents a robot architecture designed around that control problem: a 7-link arm whose six articulated joints provide twelve independently driven revolute axes, each actuated by an antagonistic pair of pneumatic muscles, so that every axis can simultaneously change its angle and linearly adjust its stiffness. The rigid articulated backbone makes the tip compliance and position of the arm predictable enough to be commanded quantitatively in real time. The robot employs a unified iterative inverse-kinematics and inverse-compliance controller to achieve simultaneous, quantitative control of both compliance and position. The task-space compliance and kinematics models and the control law are derived and verified on both the physical arm and a matched simulation. Simulation is then used to study how the same framework extends to other arm morphologies. Finally, the arm demonstrates tasks that have been difficult for both rigid and soft arms: rejecting disturbances while writing on a moving whiteboard, and passively correcting hidden misalignment during a key-insertion and drawer-opening task. That these tasks succeed under so straightforward a controller is evidence for the advantage of this algorithm-informed structural design.

Runze Zuo, Tianhua Zou, Naike Wu +2

Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI 48109, USA.

Robotic ControlImpedance ControlInverse KinematicsSoft RoboticsCompliant Robot Control
AskPDF
Sep 23, 2026

Collocated Shape Regulation for Soft Robots

Controlling the shape of a continuum soft robot typically requires an accurate dynamic model and actuation of all degrees of freedom. We show that regulating only the actuated coordinates, through collocated shape control, achieves provably stable convergence of those coordinates and, under an explicit compatibility condition, of the entire robot shape. While collocated control is a cornerstone of high-performance motion control in rigid robotics, extending this formulation to continuum soft robots has remained challenging due to the complexity of their dynamics. We present the first general framework for collocated control of continuum soft robots and derive a unified family of controllers, including PD, PID, PsatID, and their counterparts with compensation and cancellation components. The framework unifies existing approaches while introducing new controller designs. In particular, we develop three classes of PD and PID like regulators with local, semi-global, and global stability guarantees, and provide rigorous convergence analyses for each. Extensive experimental validation demonstrates the effectiveness of the proposed methods across different model discretizations and controller parameters. The resulting framework provides practical design guidelines for selecting and implementing controllers with known stability guarantees, without requiring a complete dynamic model of the robot

Pietro Pustina, Ebrahim Shahabi, Daniel Feliu-Talegon +2

Department of Computer, Control and Management Engineering, Sapienza University of Rome, Rome, Italy · Department of Cognitive Robotics, Delft University of Technology, 2628 CN Delft, The Netherlands · Institute of Robotics and Mechatronics, German Aerospace Center (DLR), 82234 Oberpfaffenhofen, Germany

Continuum Robot ControlSoft Robotics
AskPDF
Mar 23, 2026

Trajectory Generation for Underactuated Soft Robot Manipulators using Discrete Elastic Rod Dynamics

Soft robots are well suited for contact-rich tasks due to their compliance, yet this property makes accurate and tractable modeling challenging. Planning motions with dynamically-feasible trajectories requires models that capture arbitrary deformations, remain computationally efficient, and are compatible with underactuation. However, existing approaches balance these properties unevenly: continuum rod models provide physical accuracy but are computationally demanding, while reduced-order approximations improve efficiency at the cost of modeling fidelity. To address this, our work introduces a control-oriented reformulation of Discrete Elastic Rod (DER) dynamics for soft robots, and a method to generate trajectories with these dynamics. The proposed formulation yields a control-affine representation while preserving certain first-principles force-deformation relationships. As a result, the generated trajectories are both dynamically feasible and consistent with the underlying actuation assumptions. We present our trajectory generation framework and validate it experimentally on a pneumatic soft robotic limb. Hardware results demonstrate consistently improved trajectory tracking performance over a constant-curvature-based baseline, particularly under complex actuation conditions.

Beibei Liu, Akua K. Dickson, Ran Jing +1

Division of Systems Engineering, Boston University, Boston MA, USA. · Department of Mechanical Engineering, Boston University, Boston MA, USA.

Robot Motion GenerationTrajectory GenerationSoft RoboticsContinuum Robot Control
AskPDF