A novel parameter estimation method for pneumatic soft hand control applying logarithmic decrement for pseudo rigid body modeling
2025
Read A novel parameter estimation method for pneumatic soft hand control applying logarithmic decrement for pseudo rigid body modeling (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-01113.txt.
Introduction
Soft robot hands bend as they grip, which helps them handle people and fragile objects but makes control tricky. Haiyun Zhang, with four co-authors, offers a simpler way to model a soft hand driven by air pressure. Their method treats each bending part as if it were a stiff body on a hinge, a stand-in called a pseudo rigid body model.
To set the model's values, the team uses the logarithmic decrement method, which infers traits from how motion fades over time. On a test hand, their controller tracked position and force far better than a standard PID controller, with much smaller peak errors. The result supports faster, more accurate control of soft hands in real time.
Catalog record
| YEAR | 2025 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-01113
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A novel parameter estimation method for pneumatic soft hand control applying logarithmic decrement for pseudo rigid body modeling Abstract The rapid advancement in physical human-robot interaction (HRI) has propelled the growth of soft robot design and, in parallel, soft robot controllers. Controlling soft robots, especially soft hand grasping, is complex due to their ubiquitous deformation, prompting the use of reduced model-based controllers to provide sufficient state information for real-time high dynamic response control performance. However, most modeling techniques face computational efficiency and complexity of parameter identification issues and are hard to apply to real-time controls. To alleviate this, we proposed a paradigm coupling an analytical modeling approach based on Pseudo-Rigid Body Modeling and the Logarithmic Decrement Method for parameter estimation (PRBM+LDM). Using a soft robot hand test bed, we demonstrate the accuracy of PRBM+LDM to model position and force output as a function of pressure input and benchmark its performance. We, then, apply the PRBM+LDM model as a basis for a closed-loop position controller and compare its performance against a simple PID controller. Furthermore, we apply the PRBM+LDM model as a closed-loop force controller and compare its performance with simple constant pressure grasping control by performing small contact areas pinching tasks on low-weight, small objects - a screwdriver, a potato chip, and a brass coin. The PRBM+LDM-based position controller (Average Max. Error across all fingers: 4.37 \degree 4.37\degree ) outperformed the simple PID position controller (Average Max. Error across all fingers: 20.38 \degree 20.38\degree ). Furthermore, the PRBM+LDM-based force controller (Potato chip: 86%, Screwdriver: 74.42%, Brass coin: 64.75%) achieved a higher success rate than the constant pressure grasping control (Potato chip: 82.5%, Screwdriver: 70%, Brass coin: 35%) in the pinching tasks. We conclude that the PRBM+LDM modeling technique proves to be a convenient and efficient way to model the dynamic behavior of soft actuators closely and can be used to build high-precision position and force controllers. In application, it realizes stable, flexible grasping of small objects by exerting precise contact force on contact areas. Published version: This is the author-accepted manuscript of the article published in Advanced Intelligent Systems . The final version is available at Wiley Online Library: https://doi.org/10.1002/aisy.202400637 . keywords Soft Robotics, Parameter estimation, Dynamic system, Dexterous hand Haiyun Zhang* Kelvin HoLam Heung* Gabrielle J. Naquila Ashwin Hingwe Ashish D. Deshpande \dedication The first two authors, Haiyun Zhang and Kelvin HoLam Heung, contributed equally to this work. Haiyun Zhang, Gabrielle J. Naquila, Ashwin Hingwe, Ashish D. Deshpande Department of Mechanical Engineering, The University of Texas at Austin, 2515 Speedway, Austin, TX 78712, United States Email Address: [email protected]; [email protected] Kelvin HoLam Heung Department of Building and Real Estate, the Hong Kong Polytechnic University, 11 Yuk Choi Roa, Hung Hom, Hong Kong, 10587, China Email Address: [email protected] 1 Introduction The surge in physical human-robot interaction (HRI) has given rise to the necessity of compliant robotic designs to ensure safety in different human-robot workspaces [ 1 ] . This demand for compliant systems has catalyzed a wave of soft robots that provide inherent compliance and flexibility. Soft robots have been ubiquitously exploited in the last decade, and their applications have penetrated numerous robotics fields, such as surgical [ 2 , 3 , 4 , 5 ] , industrial [ 6 ] , and assistive robotics [ 7 ] , where HRI is prevalent and necessary. As the sophistication of soft robot design continues, works on soft robot controllers have also been vigorously explored in parallel. However, controlling soft robots presents unique challenges compared to rigid bodies, as soft bodies exhibit complex deformations beyond simple planar motions. Unlike rigid bodies, which can be described by six degrees of freedom (three rotations and three translations about the x, y, and z axes), soft materials are elastic and can bend, twist, stretch, compress, buckle, wrinkle, and more. This variability in motion offers an infinite number of degrees of freedom, demanding innovative approaches to modeling, control strategies, dynamics analysis, and high-level planning [ 8 ] . To attack the complexity of soft robot control, scientists have been exploring model-based and model-free controllers. Model-based controllers depend on analytical models for deriving the controller, while model-free controllers circumvent the need to use kinematic and dynamic models [ 9 ] . Historically, model-free controllers predominantly used open-loop valve sequencing to control body-segment actuation. Valve sequencing means that a valve is turned on for some duration of time to pressurize the actuator and then turned off to either hold or deflate it [ 9 ] . This type of control has also been referred to as bang-bang control [ 10 , 11 , 12 , 13 , 14 ] . In recent years, model-free controllers have been focusing on using learning-based techniques [ 15 , 16 , 17 , 18 , 19 , 20 , 21 ] . However, for most applications, model-based controls have been applicable since simplified models provide enough information to improve control performance significantly compared to the model-free baseline. Furthermore, many finite-dimensional modeling techniques have proven to be accurate, manageable, and interpretable [ 22 ] to be used over model-free controllers. Model-based static controllers are currently the most widely used and studied strategy for the control of soft robots [ 9 ] . Analytical models using Constant Curvature (CC) modeling [ 23 , 24 , 25 , 26 , 27 , 28 , 29 ] , Piecewise Constant Curvature (PCC) modeling [ 30 , 31 , 32 ] , Cosserot rod theory modeling [ 23 , 33 , 34 , 35 ] , disc-thread modeling [ 36 , 37 ] , and 3D Finite Element Models [ 38 , 39 , 40 ] have been vastly explored. However, using Cosserot rod theory and disc-thread modeling presents an overall increased computational difficulty. On the other hand, Finite Element Analysis (FEA) simulations are time-consuming, and parameter identification is dependent on tensile testing the individual materials of the actuator separately, which is also time-consuming and leads to an increase in parameter deviation from a fully assembled actuator. Furthermore, FEA assumes a quasistatic analysis which leads to further deviation from the actual behavior of the soft actuator. Across the literature, CC and PCC modeling are the most prevalent methods applied in model-based controllers due to their relative simplicity [ 9 ] . However, their non-linear analysis introduces complexity that may not be justifiable for certain applications, similar to the Cosserot rod theory and disc-thread modeling techniques. Furthermore, CC and PCC only apply when external loading effects are negligible, further constraining their applicability. For physical HRI applications, it is imperative to have a controller that can be applied conveniently, comparable to model-free methods, and can provide sufficient baseline accuracy, much like model-based controllers. Therefore, the computational efficiency of the model, the convenience of applying the modeling technique, and the ability to properly model the dynamics of the soft actuator under considerable external load and under a non-quasistatic state are crucial for effective control system development. Hence, in this work, we employ the Pseudo-Rigid Body Modeling (PRBM), introduced in [ 41 ] and [ 42 ] , to build the dynamic equation for model-based control of soft robots and complement it with the Logarithmic Decrement (LDM) technique for parameter