8 July 2024 in Advanced Intelligent Systems
Soft robots can adaptively interact with unstructured environments. However, nonlinear soft material properties challenge modeling and control. Learning‐based controllers that leverage efficient mechanical models are promising for solving complex interaction tasks. This article develops a closed‐loop pose/force controller for a dexterous soft manipulator enabling dynamic pushing tasks using deep reinforcement learning. Force tests investigate the mechanical properties of a soft robot module, resulting in orthogonal forces of N. Then, the policy is trained in simulation leveraging a dynamic Cosserat rod model of the soft robot. Domain randomization mitigate the sim‐to‐real gap while careful reward engineering induced pose and force control even without explicit force inputs. Despite the approximate simulation, the sim‐to‐real transfer achieved an average reaching distance of mm (), an average orientation error of rad () and applied pushing forces up to N. Such performance is reasonable for the intended assistive tasks of the manipulator. The experiments uncovered that the soft robot interacting with the environment exhibited torsional and counter‐balancing movements. Although not explicitly enforced, they emerged from the mechanical intelligence of the manipulator. The results demonstrate the potential of soft robotic manipulation via reinforcement learning.
29 September 2023 in IEEE Robotics & Automation Magazine
In this article, we present, for the first time, a soft robot control system (SofToss) capable of throwing life-size objects toward target positions. SofToss is an open-loop controller based on deep reinforcement learning (RL) that generates, given the target position, an actuation pattern for the tossing task. To deal with the high nonlinearity of the dynamics of soft robots, we deploy a neural network to learn the relationship between the actuation pattern and the target landing position, i.e., the direct model (DM) of the task. Then, an RL method is used to predict the actuation pattern given the goal position. The proposed controller was tested on a modular soft robotic arm, I-Support, by tossing four objects of different shapes and weights in 140-mm squared target boxes. We registered a success rate of almost 65% of the throws in two actuation modalities (i.e., partial, keeping one module of the soft arm passive, and complete, with both modules active). This performance raises to 85% if one can choose the number of modules to actuate for each throwing direction. Furthermore, the results show that the proposed learning-based, real-time controller achieves a performance comparable to that of an optimization-based, nonreal-time controller. Our study contributes to the foundations for bringing soft robots into everyday life and industry by performing more complex, dynamic tasks.
7 November 2024 in Advanced Intelligent Systems
Soft manipulators, renowned for their compliance and adaptability, hold great promise in their ability to engage safely and effectively with intricate environments and delicate objects. Nonetheless, controlling these soft systems presents distinctive hurdles owing to their nonlinear behavior and complicated dynamics. Learning‐based controllers for continuum soft manipulators offer a viable alternative to model‐based approaches that may struggle to account for uncertainties and variability in soft materials, limiting their effectiveness in real‐world scenarios. Learning‐based controllers can be trained through experience, exploiting various forward models that differ in physical assumptions, accuracy, and computational cost. In this article, the key features of popular forward models, including geometrical, pseudo‐rigid, continuum mechanical, or learned, are first summarized. Then, a unique characterization of learning‐based policies, emphasizing the impact of forward models on the control problem and how the state of the art evolves, is offered. This leads to the presented perspectives outlining current challenges and future research trends for machine‐learning applications within soft robotics.
13 August 2024 in IEEE Robotics and Automation Letters
In this letter, we present SoftSling, a soft robot control strategy designed for accurately throwing objects following circular run-ups. SoftSling draws inspiration from ancient slingers, who rotated a sling loaded with a projectile at high speeds to fight and hunt, releasing the object by letting go of the sling's end. Our study aims to replicate this behavior by exploiting the embodied intelligence of soft robots under periodic actuation input, that enables them to generate self-stabilizing motions. The periodic input parameters for moving along a circular-like path are generated by a neural network based on the weight of the object and the target position. Subsequently, a separate neural network model predicts the release time by considering the gripper opening delay and the object positions during motion. We tested this strategy on a modular soft robot, I-Support, by throwing three objects of varying weights into 140-mm square target boxes. We achieved a success rate ranging from 75% to 88% for different objects, with the heaviest object yielding the highest success rate. Our research contributes to integrating soft robots into everyday life, enabling them to perform complex and dynamic tasks.
23 March 2026 in IEEE Access
This work presents a novel control strategy for the hand-guidance of a soft continuum arm. This promotes human-robot interaction in collaborative robotics; the proposed control system is implemented on a modular soft manipulator, exploiting its intrinsic safety. During hand guidance, users exert forces on the robot’s end effector, which dynamically generates a smooth path for the robot to follow. A learning-based model is employed to predict the actuation patterns required to achieve desired positions, effectively solving the robot’s inverse dynamics. The system’s performance was evaluated under four distinct scenarios. In the human-guidance usability test, the controller demonstrated an average position error of 7.73 ± 0.96 mm. To validate robustness across different users, an inter-subject variability test with 10 participants yielded a consistent average error of 6.24 ± 3.05 mm. The system’s resilience was further assessed in a noisy environment test, where the error remained bounded between 6.99 ± 3.40 mm and 34.64 ± 14.75 mm even under severe sensor degradation. Finally, under motor-driven guidance, the controller ensured accuracy across varying movement directions with an average position error of 3.82±3.32 mm. These results underline the potential of soft robots for safe and interactive use in everyday environments, showcasing their suitability for collaborative applications.
27 August 2026 in Advanced Intelligent Systems
The elephant trunk is a dexterous and versatile manipulator whose performance is still unmatched in robotics. In previous works,modularity was prioritized andrelatively small‐scale continuum robots were built. We take the natural proboscis of the Loxodonta africana species as a model and propose a different design approach which favors structural continuity andnatural frequencythatplausibly emulate those of the natural trunk, while conferring high adaptability to the environment and humans. Instead of targeting prescribed behaviors, we show that a biomimetic design based on the macroscopic properties of the natural system enables elephant‐like movements and grasping. We build by 3D printing an 85 cm long, compliant, tapered, volumetrically tessellated continuum arm, which is combined with tendon‐driven actuation mimicking the longitudinal and oblique muscles of the natural model. We demonstrate whole‐body grasping of objects having different shapes and dimensions and discuss a comparison to the biological trunk highlighting aspects of both biology and robotics.