Multi-Modular MANTA-RAY: A Modular Soft Surface Platform for Distributed Multi-Object Manipulation
Pratik Ingle, Jørn Lambertsen, Kasper Støy, Andres Faiña · 2026
Read Multi-Modular MANTA-RAY: A Modular Soft Surface Platform for Distributed Multi-Object Manipulation by Pratik Ingle, Jørn Lambertsen, Kasper Støy, Andres Faiña (2026) 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-2601-21884.txt.
Introduction
Pratik Ingle, Jørn Lambertsen, Kasper Støy, and Andres Faiña build surfaces that move objects without grabbing them. A manipulation surface controls items by deforming its own shape, tipping and rolling things across itself. Dense arrays of actuators can make rich shapes, but they bring many degrees of freedom and hard control problems.
The team's MANTA-RAY platform takes a modular route. And this 2026 preprint describes a multi-modular version for handling several objects at once in a distributed way. Soft, modular surfaces could one day sort fragile goods. They could feed parts to assembly robots. They could move items where grippers are too clumsy. All this works by spreading control across simple repeating units.
Catalog record
| CREATOR | Pratik Ingle, Jørn Lambertsen, Kasper Støy, Andres Faiña |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-21884
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Multi-Modular MANTA-RAY: A Modular Soft Surface Platform for Distributed Multi-Object Manipulation Pratik Ingle, Jørn Lambertsen, Kasper Støy, Andres Faiña Abstract Manipulation surfaces control objects by actively deforming their shape rather than directly grasping them. While dense actuator arrays can generate complex deformations, they also introduce high degrees of freedom (DOF), increasing system complexity and limiting scalability. The MANTA-RAY (Manipulation with Adaptive Non-rigid Textile Actuation with Reduced Actuation densitY) platform addresses these challenges by leveraging a soft, fabric-based surface with reduced actuator density to manipulate fragile and heterogeneous objects. Previous studies focused on single-module implementations supported by four actuators, whereas the feasibility and benefits of a scalable, multi-module configuration remain unexplored. In this work, we present a distributed modular and scalable variant of the MANTA-RAY platform that maintains manipulation performance with a reduced actuator density. The proposed multi-module MANTA-RAY platform and control strategy employs a object passing between modules and geometric transformation-driven PID controller that directly maps tilt-angle control outputs to actuator commands, eliminating the need for extensive data-driven or black-box training. We evaluate system performance in simulation across surface configurations of varying modules ( 3 × 3 3\times 3 , and 4 × 4 4\times 4 ) and validate its feasibility through experiments on a physical 2 × 2 2\times 2 hardware prototype. The system successfully manipulates objects with diverse geometries, masses, and textures—including fragile items such as eggs and apples as well enabling parallel manipulation. The results demonstrate that the muti-module MANTA-RAY improves scalability and enables coordinated manipulation of multiple objects across larger areas, highlighting its potential for practical, real-world applications. Index Terms – Manipulation, Soft Robot, Robotic Manipulation Surface, Distributed I Introduction Fig. 1 : Multi-Modular MANTA-RAY: Manipulation with Adaptive Non-rigid Textile Actuation with Reduced Actuation densitY platform. (A) Manipulation of an egg and an apple. (B) Side view showing the platform dimensions of 1\times 1\times 0.7 m (length, width, and height at rest). (C) Top view illustrating the arrangement of nine actuators (A0–A8) and four modules (M0–M3) without the flexible surface. (D) Each linear actuator has 0.4-meter vertical range and built using a stepper motor and pulley-belt mechanism, controlled by an Arduino Uno micro controller. (E) MuJoCo simulation of the extended 3\times 3 -module configuration showing two objects (red and blue spheres) and their respective target locations. Objects in real-world environments exhibit wide variations in shape, size, and material properties, making it difficult for robotic manipulators to handle them. Conventional grippers, though precise for structured tasks, often fail with irregular or fragile items—too little force leads to slipping, while too much can cause damage. These limitations highlight the need for adaptable, contact-rich manipulation strategies that can generalize across diverse object types. In recent years Robotic Manipulation Surfaces (RMS) have emerged as a promising approach to address these challenges. Rather than grasping objects directly, RMS platforms manipulate them by dynamically deforming or actuating the surface on which the objects rest. Arrays of actuators generate local deformations or motions that leverage friction, gravity, and vibration to reposition one or multiple objects simultaneously. The distributed contact area provided by such surfaces allows safe and coordinated manipulation, particularly advantageous for fragile or large, flat objects that benefit from uniform support. Several RMS designs been proposed over the time. Wheel-based systems [ 7 , 10 ] use omnidirectional rollers to transport items efficiently but struggle with small or uneven objects that cannot maintain stable contact. Piston-based arrays [ 12 , 5 ] offer precise local control by vertically actuating dense grids of pins, enabling translation and rotation. However, their reliance on high actuator density increases mechanical complexity, cost, and control overhead. Other approaches—such as vibrating plates [ 2 , 14 ] , air-flow levitation [ 6 ] , cilia-inspired actuators [ 1 ] , and deformable gel layers [ 8 ] —demonstrate unique advantages but remain limited by trade-offs in payload, versatility, or scalability. To address these challenges, Chih-Han et al. [ 13 ] proposed Morpho , a self-deformable modular robot inspired by cellular structures that enables large-scale shape morphing for diverse global functions. Using neoprene and latex membranes, it demonstrates potential for surface-based object manipulation but focuses primarily on inter-module ball transfer rather than precise manipulation. Recently Ingle et al. proposed MANTA-RAY: Manipulation with Adaptive Non-rigid Textile Actuation with Reduced Actuation densitY [ 4 ] . A novel system that drastically reduces the hardware and control complexity. It achieves heterogeneous and fragile object manipulation over a large area with only a few actuators. It uses a flexible fabric sheet attached to a small number of linear actuators forming a reconfigurable surface that can tilt and deform in order to move objects indirectly. By interconnecting the actuators with a continuous soft layer, the MANTA-RAY effectively uses the surface itself as a medium to transmit forces to the object, enabling even small or irregular items to be manipulated without falling through gaps. This approach dramatically lowers the actuator count and degrees of freedom [ 11 ] compared to traditional RMS. The reduced hardware not only cuts down on system cost and weight, but also simplifies control since there are far fewer motors to coordinate. Even with fewer actuators, MANTA-RAY preserves versatile manipulation performance [ 3 ] , as its compliant fabric surface conforms to object geometry and safely handles fragile items—including eggs and fruits—without damage. Notably, the system achieves an object-to-actuator size ratio as low as 0.01 (i.e., objects can be 100\times smaller than the spacing between actuators) – far beyond what previous manipulation surfaces could achieve. Despite its advantages, heterogeneous object manipulation has so far been demonstrated only on a single MANTA-RAY module, where the surface is supported and controlled by exactly four actuators. The potential benefits and feasibility of extending this design into a multi-modular configuration—where multiple actuators are arranged in a grid to form several independent modules, each supported by four actuators—remain largely unexplored. This limitation primarily arises from the absence of an effective multi-module control framework and compatible hardware design. A multi-modular MANTA-RAY architecture, however, introduces the possibility of parallel manipulation , enabling the simultaneous handling of multiple heterogeneous objects across interconnected modules. In this work, we propose a multi-modular, distributed MANTA-RAY platform capable of manipulating heterogeneous objects and achieving parallel, coordinated manipulation across multiple modules. In the hardware implementation, each actuator functions as an independent unit, enhancing the system’s robustness to individual actuator faults. The modular design further allows adjustable actuator spacing to accommodate various workspace dimensions and application requirements. We demonstrate that a classical linear controller, integrated with an object-passing strategy, can effectively control this inherently soft and non-linear system when combined with an appropriate geometric transformation. This approach enables seamless object transfer between neighboring modules, allowing continuous, robust, and efficient manipulation across the entire surface. II Multi-Modular MANTA-RAY The single-module MANTA-RAY platform manipulates objects on a soft fabric surface using four linear actuators placed at the corners of a square and that move vertically and change the slope of the fabric. To extend its scalability, we developed a 3\times 3 actuator grid configuration, forming a 2\times 2 multi-modular MANTA-RAY platform (see Fig. 1 ). Each module (M0–M3) consists of four actuators selected from the nine available actuators (A0–A8), as shown in Fig. 1 C. Adjacent modules