Development of Multi-chamber Pneumatic Twist Actuator for Soft robot

Soft-robotics is a sub-branch of robotics that deals with breaking this traditional perspective of rigid robots by making the robot more compliant and inspired by nature. Conventional electromechanical actuators are precise in operation but they are rigid and heavy. To make a robot inherently soft there is need to make soft actuators which are compliant. Various types of Soft actuators were previously developed for the bending, contraction and expansion purpose. In this following section, we are describing the development of the twist actuator made up of PDMS (Polydimethylsiloxane). we did the dynamic simulation of the actuator in ANSYS by using hyper-elastic material model and also fabricated complex shape core of actuator using a PDMS casting process in sacrificial 3d printed moulds. we tested these actuators for twisting rotation, torque and actuation time with suitable air pressure and also validated the results of simulation by experiment in a test setup.

Skills Used are Solidworks, Finite Element Analysis (ANSYS), Hyperelastic Material Model, Design of sacrificial molds, Casting, Electronics to build test setup, Rapid Prototyping, Arduino etc


The axial rotation about the Z axis is actuated using twist actuator. It works on the principle of inflation of the pneumatic chamber. There are 7 pneumatic chambers, wound coaxially and helically. On the application of pressurised air, individual chamber inflates and tries to unwound  As the extreme end are constrained, we get a net rotation (as well as a small extension in Z direction) along the Z-axis. Some important design variables are, material properties, chamber profile design, helix angle, length of helix, this section discusses the design, investigation of various design parameters and the simulation of the twist actuator. The geometry was designed in SolidWorks and is as shown in Fig 1.

                                                                                                                 Figure 1 a)Geometry of Twist Actuator

The internal walls have applied a pressure of 0.1 MPa (Ramped) gauge. The lower surface is fixed. Entire simulation setup is shown in Fig. 2

                                                                                                                 Figure 2 Cross section of Twist Actuator

                                                             Figure 3 Simulation setup of Twist Actuator        Figure 4 Mesh convergence of Twist Actuator

The features in control volume were meshed using quadratic triangular/ tetrahedron elements as large deformations are expected. The quadratic elements capture curved surface deformations and relatively larger size elements can be used as compared to linear elements for same computation capacity and accuracy. The Meshing is shown in Figure 4. An iterative solver was selected due to the quasi-static nature of the simulation. The results are shown in Fig. 5

                                                                 5) a)Directional Deformation of Twist Actuator b)DirectionalDeformationTop View

The angular displacement was calculated geometrically from the directional deformation contour of the actuator as:

Where Y is the maximum deformation in the Y direction and is the rotation. The results are summarised in the table below

Table 1 Variation of Twist angle for different Pressure Values

Pressure (MPa gauge)








                                                                                           Figure 6Twist Actuator total deformation cross-sectional view

                                                                                                              Figure 7) Twist Actuator Analysis result

It can be inferred that as the operating pressure increases, the radius of curvature of each membrane wall will decrease and thus the stroke will increase. It is also a function of no. of chambers and the winding angle (which is 180o).  The torque has a positive coupling with the chamber wall surface area normal to the axial direction of an actuator and the operating pressure. The simulation needs to be done in Time domain to reduce the significance of the mass flow rate assumption.

The free rotating surface was restrained, and the simulation was rerun, the support moment reaction was indicative of torque produced by the actuator. The simulation results are summarized in table below.

Table  2 Variation of Torque for different Pressure Values

Pressure (MPa gauge)

 (Deg) (kgcm)







The Soft actuators are currently designed assuming time constant loads and forces. However, a time domain investigation and damping characteristic determination and improvements are yet to be done. Upon a thorough material property investigation, it is a pretty straightforward problem. During Simulation, the defects occurring during the manufacturing (which are abundant) were neglected. This has produced certain inaccuracies in simulated results and actual results.


Conventional twist actuators contain motors, made up of a heavy metal structure and thus have a heavy mass. Designing a lightweight Soft twist actuator was a challenge in itself. The design of the twist actuator comprises of a complex helical structure that has helical pockets in it that untwist themselves when pressurized air is injected into them.

Initially, for the first iteration, a downsized twist actuator was initially manufactured. Its mold consisted of an Inner core part and an Outer Mold. The first iteration of the actuator was designed to have a wall thickness of 2.5mm and thus the inner core and the outer mold had a very small pouring gap between them. Hence, No proper formation and casting of the helix were achieved. For the second iteration, the wall thickness was hence increased to 5mm in order to make the material pouring process easier. For this optimized iteration, The size and design of the actuator and shoulder size were also subsequently modified.

Similar to the elbow actuator mold the inner core of the twist actuator mold as also sacrificial in nature because of its intricate shape and size. Also, since the mold was very intricate in nature, Using HT45 PDMS, that had very small working time, would have been a failure because of the inability of the material to flow into all the parts of the mold. Hence, HT33 was the material selected, which has a relatively higher working time and higher flexibility for easy untwisting while actuation.

From the analysis of the design of the second iteration, The solution converged to get a twist of 28 degrees. Actually, after testing the twist actuator, a twist of 36 degrees was achieved.

Two different kinds of helixes (one left hand and one right hand) was required for the left hand and the right hand respectively. The whole cured part along with the inner and outer mold was immersed to dissolve into Acetone overnight.

The HT33 helix structure was later joint with two HT33 connectors, which were subsequently attached to two more HT45 connectors. Hence, the twist actuator part was a 5 stage intricate mold.

The joining of the helix to the connector was difficult for sealing, thus, for the third iteration of the twist actuator, a cylindrical extrusion was given to the helix and it was easily able to be joined to the connectors.

 Thus, A soft helical pneumatic twist actuator was manufactured.

                                                                           Figure 8) a)Twist Actuator first Prototype              b)Twist Actuator split mold

                                                                           Figure 9 a)Right and Left Helix Twist Actuator Inner core and Outer core mold

                                                                          Figure 10  a) Right Helix b)After casting of Twist Actuator c)Mold dipped in acetone

                                                                                                  Figure 11  Sacrificial PLA after removed from casting


The test setup of the twist actuator was necessary to be different in nature because the twisting torque was to be measured tangentially to the curved surface of the actuator. As shown in the following figures, a test setup was made by constraining the base of the twist actuator and attaching a rigid force transmitting link to transmit the force through an elastic string to a vertically mounted cantilever load cell. The load cell was programmed and calibrated before testing. The readings of maximum torque obtained at different constant pressure actuation were noted and plotted. Also, the actuation cycle times of the actuator for a twist of 45 degrees were measured within a range of pressure between 1 to 3 bar with steps of 0.5 bar. The results obtained from the tests were plotted as follows.

                                                                                                                 Figure 12 Twist actuator rotation test 


                                                                                                              Figure 13 Twist Actuator testing Setup

                                                                                     Figure 14 Twist Actuator Actuation time vs Pressure for 45 degree twist

Test results for twist actuator are plotted as shown in the graph (figure 9.9), on the abscissa pressure is varying from 0 bar to 3 bar and on ordinate actuation time to get 45degree twist is indicated. Actuation time is more for the low pressure and as pressure increases actuation time decreases. The angular velocity of the Arm is varying in between 0.12 rad/sec to .46 rad/sec at pressure 1 bar to 3 bar respectively.

                                                                                                              Figure 15 Twist Actuator Torque vs Pressure      

Test results for twist actuator are plotted as shown in the graph (figure 9.10), on the abscissa pressure is varying from 0 bar to 3.5 bar and on ordinate torque in the Kgcm is indicated. As the pressure increases the torque is increasing linearly maximum 6.9 Kgcm torque is obtained by the twist actuator at the pressure 3.5 bar.