Cosimo Della Santina

Zhaoting Li, Gang Chen, Javier Alonso-Mora et al.

Diffusion policies have recently emerged as a powerful framework for robotic manipulation. However, like other behavior cloning methods, they remain vulnerable to distributional shift, often requiring human-in-the-loop interventions to correct failures during deployment. These interactions naturally provide paired supervision in the form of the robot's undesired actions and the human teacher's corrective actions. Yet existing data aggregation pipelines and standard behavior cloning losses largely ignore this negative signal from undesired actions, leading to overfitting to teacher's actions and an increasing reliance on costly expert data. To address this limitation, we propose Set-Supervised Diffusion Policy (SDP), a novel learning framework that utilizes contrastive action-chunk data to train diffusion policies from human corrections. From paired positive and negative action-chunks, SDP constructs a set of desired action-chunks and designs a training pipeline that encourages the diffusion policy to align with the set. Through extensive experiments across multiple robotic manipulation tasks, we demonstrate that SDP consistently improves policy performance, with particularly strong gains in robustness to noisy data. Moreover, SDP induces high-quality aggregated datasets, enabling more efficient and reliable policy learning from human-in-the-loop corrections. Our code is available at https://set-supervised-diffusion-policy.github.io/.

Zhaoting Li, Rodrigo Pérez-Dattari, Robert Babuska et al.

Behavior cloning (BC) optimizes policies by treating human demonstrations as pointwise action labels. While effective with accurate action labels, this formulation is brittle in practice: when human-provided actions are imperfect, treating each label as an exact target can steer the policy away from the underlying desired behavior, particularly when expressive models are used (e.g., energy-based models). As a result, we propose a human-in-the-loop alternative that replaces pointwise supervision with set-valued action targets. We introduce Contrastive policy Learning from Interactive Corrections (CLIC). CLIC leverages human corrections to construct and refine sets of desired actions, and optimizes a policy to place probability mass over these sets rather than over a single action target. This formulation naturally accommodates both absolute and relative corrections and can represent complex multi-modal behaviors. Extensive simulation and real-robot experiments show that the proposed approach leads to effective policy learning across diverse settings: CLIC remains competitive with the state of the art under accurate data while being substantially more robust under noisy, relative, and partial feedback. Our implementation is publicly available at https://clic-webpage.github.io/.

Yulei Qiu, Jihong Zhu, Cosimo Della Santina et al.

Deformable Object Manipulation (DOM) is an important field of research as it contributes to practical tasks such as automatic cloth handling, cable routing, surgical operation, etc. Perception is considered one of the major challenges in DOM due to the complex dynamics and high degree of freedom of deformable objects. In this paper, we develop a novel image-processing algorithm based on Gabor filters to extract useful features from cloth, and based on this, devise a strategy for cloth flattening tasks. We also evaluate the overall framework experimentally and compare it with three human operators. The results show that our algorithm can determine the direction of wrinkles on the cloth accurately in simulation as well as in real robot experiments. Furthermore, our dewrinkling strategy compares favorably to baseline methods. The experiment video is available on https://sites.google.com/view/robotic-fabric-flattening/home

Maximilian Stölzle, Cosimo Della Santina

Even though a variety of methods have been proposed in the literature, efficient and effective latent-space control (i.e., control in a learned low-dimensional space) of physical systems remains an open challenge. We argue that a promising avenue is to leverage powerful and well-understood closed-form strategies from control theory literature in combination with learned dynamics, such as potential-energy shaping. We identify three fundamental shortcomings in existing latent-space models that have so far prevented this powerful combination: (i) they lack the mathematical structure of a physical system, (ii) they do not inherently conserve the stability properties of the real systems, (iii) these methods do not have an invertible mapping between input and latent-space forcing. This work proposes a novel Coupled Oscillator Network (CON) model that simultaneously tackles all these issues. More specifically, (i) we show analytically that CON is a Lagrangian system - i.e., it possesses well-defined potential and kinetic energy terms. Then, (ii) we provide formal proof of global Input-to-State stability using Lyapunov arguments. Moving to the experimental side, we demonstrate that CON reaches SoA performance when learning complex nonlinear dynamics of mechanical systems directly from images. An additional methodological innovation contributing to achieving this third goal is an approximated closed-form solution for efficient integration of network dynamics, which eases efficient training. We tackle (iii) by approximating the forcing-to-input mapping with a decoder that is trained to reconstruct the input based on the encoded latent space force. Finally, we show how these properties enable latent-space control. We use an integral-saturated PID with potential force compensation and demonstrate high-quality performance on a soft robot using raw pixels as the only feedback information.

Eveline Drijver, Rodrigo Pérez-Dattari, Jens Kober et al.

Intelligent manufacturing is becoming increasingly important due to the growing demand for maximizing productivity and flexibility while minimizing waste and lead times. This work investigates automated secondary robotic food packaging solutions that transfer food products from the conveyor belt into containers. A major problem in these solutions is varying product supply which can cause drastic productivity drops. Conventional rule-based approaches, used to address this issue, are often inadequate, leading to violation of the industry's requirements. Reinforcement learning, on the other hand, has the potential of solving this problem by learning responsive and predictive policy, based on experience. However, it is challenging to utilize it in highly complex control schemes. In this paper, we propose a reinforcement learning framework, designed to optimize the conveyor belt speed while minimizing interference with the rest of the control system. When tested on real-world data, the framework exceeds the performance requirements (99.8% packed products) and maintains quality (100% filled boxes). Compared to the existing solution, our proposed framework improves productivity, has smoother control, and reduces computation time.

Peiyu Yang, Jiatao Ding, Wei Pan et al.

Achieving safe quadrupedal locomotion in real-world environments has attracted much attention in recent years. When walking over uneven terrain, achieving reliable estimation and realising safety-critical control based on the obtained information is still an open question. To address this challenge, especially for low-cost robots equipped solely with proprioceptive sensors (e.g., IMUs, joint encoders, and contact force sensors), this work first presents an estimation framework that generates a 2.5-D terrain map and extracts support plane parameters, which are then integrated into contact and state estimation. Then, we integrate this estimation framework into a safety-critical control pipeline by formulating control barrier functions that provide rigorous safety guarantees. Experiments demonstrate that the proposed terrain estimation method provides smooth terrain representations. Moreover, the coupled estimation framework of terrain, state, and contact reduces the mean absolute error of base position estimation by 64.8%, decreases the estimation variance by 47.2%, and improves the robustness of contact estimation compared to a decoupled framework. The terrain-informed CBFs integrate historical terrain information and current proprioceptive measurements to ensure global safety by keeping the robot out of hazardous areas and local safety by preventing body-terrain collision, relying solely on proprioceptive sensing.

Vito Daniele Perfetta, Daniel Feliu Talegon, Ebrahim Shahabi et al.

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.

Petr Vanc, Giovanni Franzese, Jan Kristof Behrens et al.

Learning from demonstration is a promising approach for teaching robots new skills. However, a central challenge in the execution of acquired skills is the ability to recognize faults and prevent failures. This is essential because demonstrations typically cover only a limited set of scenarios and often only the successful ones. During task execution, unforeseen situations may arise, such as changes in the robot's environment or interaction with human operators. To recognize such situations, this paper focuses on teaching the robot situational awareness by using a camera input and labeling frames as safe or risky. We train a Gaussian Process (GP) regression model fed by a low-dimensional latent space representation of the input images. The model outputs a continuous risk score ranging from zero to one, quantifying the degree of risk at each timestep. This allows for pausing task execution in unsafe situations and directly adding new training data, labeled by the human user. Our experiments on a robotic manipulator show that the proposed method can reliably detect both known and novel faults using only a single example for each new fault. In contrast, a standard multi-layer perceptron (MLP) performs well only on faults it has encountered during training. Our method enables the next generation of cobots to be rapidly deployed with easy-to-set-up, vision-based risk assessment, proactively safeguarding humans and detecting misaligned parts or missing objects before failures occur. We provide all the code and data required to reproduce our experiments at imitrob.ciirc.cvut.cz/publications/ilesia.

Rodrigo Pérez-Dattari, Zhaoting Li, Robert Babuška et al.

Planning methods struggle with computational intractability in solving task-level problems in large-scale environments. This work explores leveraging the commonsense knowledge encoded in LLMs to empower planning techniques to deal with these complex scenarios. We achieve this by efficiently using LLMs to prune irrelevant components from the planning problem's state space, substantially simplifying its complexity. We demonstrate the efficacy of this system through extensive experiments within a household simulation environment, alongside real-world validation using a 7-DoF manipulator (video https://youtu.be/6ro2UOtOQS4).

Marco Lepri, Davide Bacciu, Cosimo Della Santina

This work concerns control-oriented and structure-preserving learning of low-dimensional approximations of high-dimensional physical systems, with a focus on mechanical systems. We investigate the integration of neural autoencoders in model order reduction, while at the same time preserving Hamiltonian or Lagrangian structures. We focus on extensively evaluating the considered methodology by performing simulation and control experiments on large mass-spring-damper networks, with hundreds of states. The empirical findings reveal that compressed latent dynamics with less than 5 degrees of freedom can accurately reconstruct the original systems' transient and steady-state behavior with a relative total error of around 4\%, while simultaneously accurately reconstructing the total energy. Leveraging this system compression technique, we introduce a model-based controller that exploits the mathematical structure of the compressed model to regulate the configuration of heavily underactuated mechanical systems.

Ricardo Valadas, Maximilian Stölzle, Jingyue Liu et al.

Obtaining dynamic models of continuum soft robots is central to the analysis and control of soft robots, and researchers have devoted much attention to the challenge of proposing both data-driven and first-principle solutions. Both avenues have, however, shown their limitations; the former lacks structure and performs poorly outside training data, while the latter requires significant simplifications and extensive expert knowledge to be used in practice. This paper introduces a streamlined method for learning low-dimensional, physics-based models that are both accurate and easy to interpret. We start with an algorithm that uses image data (i.e., shape evolutions) to determine the minimal necessary segments for describing a soft robot's movement. Following this, we apply a dynamic regression and strain sparsification algorithm to identify relevant strains and define the model's dynamics. We validate our approach through simulations with various planar soft manipulators, comparing its performance against other learning strategies, showing that our models are both computationally efficient and 25x more accurate on out-of-training distribution inputs. Finally, we demonstrate that thanks to the capability of the method of generating physically compatible models, the learned models can be straightforwardly combined with model-based control policies.

Gioele Buriani, Jingyue Liu, Maximilian Stölzle et al.

Reduced-order models are essential for motion planning and control of quadruped robots, as they simplify complex dynamics while preserving critical behaviors. This paper introduces a novel methodology for deriving such interpretable dynamic models, specifically for jumping. We capture the high-dimensional, nonlinear jumping dynamics in a low-dimensional latent space by proposing a learning architecture combining Sparse Identification of Nonlinear Dynamics (SINDy) with physical structural priors on the jump dynamics. Our approach demonstrates superior accuracy to the traditional actuated Spring-loaded Inverted Pendulum (aSLIP) model and is validated through simulation and hardware experiments across different jumping strategies.

Chuhan Zhang, Cong Wang, Wei Pan et al.

Inspired by the dynamic coupling of moto-neurons and physical elasticity in animals, this work explores the possibility of generating locomotion gaits by utilizing physical oscillations in a soft snake by means of a low-level spiking neural mechanism. To achieve this goal, we introduce the Double Threshold Spiking neuron model with adjustable thresholds to generate varied output patterns. This neuron model can excite the natural dynamics of soft robotic snakes, and it enables distinct movements, such as turning or moving forward, by simply altering the neural thresholds. Finally, we demonstrate that our approach, termed SpikingSoft, naturally pairs and integrates with reinforcement learning. The high-level agent only needs to adjust the two thresholds to generate complex movement patterns, thus strongly simplifying the learning of reactive locomotion. Simulation results demonstrate that the proposed architecture significantly enhances the performance of the soft snake robot, enabling it to achieve target objectives with a 21.6% increase in success rate, a 29% reduction in time to reach the target, and smoother movements compared to the vanilla reinforcement learning controllers or Central Pattern Generator controller acting in torque space.

Chuhan Zhang, Wei Pan, Cosimo Della Santina

Motor imagery, an important category in electroencephalogram (EEG) research, often intersects with scenarios demanding low energy consumption, such as portable medical devices and isolated environment operations. Traditional deep learning algorithms, despite their effectiveness, are characterized by significant computational demands accompanied by high energy usage. As an alternative, spiking neural networks (SNNs), inspired by the biological functions of the brain, emerge as a promising energy-efficient solution. However, SNNs typically exhibit lower accuracy than their counterpart convolutional neural networks (CNNs). Although attention mechanisms successfully increase network accuracy by focusing on relevant features, their integration in the SNN framework remains an open question. In this work, we combine the SNN and the attention mechanisms for the EEG classification, aiming to improve precision and reduce energy consumption. To this end, we first propose a Non-iterative Leaky Integrate-and-Fire (LIF) neuron model, overcoming the gradient issues in the traditional SNNs using the Iterative LIF neurons. Then, we introduce the sequence-based attention mechanisms to refine the feature map. We evaluated the proposed Non-iterative SNN with Attention (NiSNN-A) model on OpenBMI, a large-scale motor imagery dataset. Experiment results demonstrate that 1) our model outperforms other SNN models by achieving higher accuracy, 2) our model increases energy efficiency compared to the counterpart CNN models (i.e., by 2.27 times) while maintaining comparable accuracy.

Semanur Küçük, Cosimo Della Santina, Angeliki Laskari

Segmenting gas bubbles in multiphase flows is a critical yet unsolved challenge in numerous industrial settings, from metallurgical processing to maritime drag reduction. Traditional approaches-and most recent learning-based methods-assume near-spherical shapes, limiting their effectiveness in regimes where bubbles undergo deformation, coalescence, or breakup. This complexity is particularly evident in air lubrication systems, where coalesced bubbles form amorphous and topologically diverse patches. In this work, we revisit the problem through the lens of modern vision foundation models. We cast the task as a transfer learning problem and demonstrate, for the first time, that a fine-tuned Segment Anything Model SAM v2.1 can accurately segment highly non-convex, irregular bubble structures using as few as 100 annotated images.

Maximilian Stölzle, T. Konstantin Rusch, Zach J. Patterson et al. · eth-zurich

Learning from demonstration provides a sample-efficient approach to acquiring complex behaviors, enabling robots to move robustly, compliantly, and with fluidity. In this context, Dynamic Motion Primitives offer built - in stability and robustness to disturbances but often struggle to capture complex periodic behaviors. Moreover, they are limited in their ability to interpolate between different tasks. These shortcomings substantially narrow their applicability, excluding a wide class of practically meaningful tasks such as locomotion and rhythmic tool use. In this work, we introduce Orbitally Stable Motion Primitives (OSMPs) - a framework that combines a learned diffeomorphic encoder with a supercritical Hopf bifurcation in latent space, enabling the accurate acquisition of periodic motions from demonstrations while ensuring formal guarantees of orbital stability and transverse contraction. Furthermore, by conditioning the bijective encoder on the task, we enable a single learned policy to represent multiple motion objectives, yielding consistent zero-shot generalization to unseen motion objectives within the training distribution. We validate the proposed approach through extensive simulation and real-world experiments across a diverse range of robotic platforms - from collaborative arms and soft manipulators to a bio-inspired rigid-soft turtle robot - demonstrating its versatility and effectiveness in consistently outperforming state-of-the-art baselines such as diffusion policies, among others.

Saray Bakker, Rodrigo Pérez-Dattari, Cosimo Della Santina et al.

Using the language of dynamical systems, Imitation learning (IL) provides an intuitive and effective way of teaching stable task-space motions to robots with goal convergence. Yet, IL techniques are affected by serious limitations when it comes to ensuring safety and fulfillment of physical constraints. With this work, we solve this challenge via TamedPUMA, an IL algorithm augmented with a recent development in motion generation called geometric fabrics. As both the IL policy and geometric fabrics describe motions as artificial second-order dynamical systems, we propose two variations where IL provides a navigation policy for geometric fabrics. The result is a stable imitation learning strategy within which we can seamlessly blend geometrical constraints like collision avoidance and joint limits. Beyond providing a theoretical analysis, we demonstrate TamedPUMA with simulated and real-world tasks, including a 7-DoF manipulator.

Cosimo Della Santina, Christian Duriez, Daniela Rus

Continuum soft robots are mechanical systems entirely made of continuously deformable elements. This design solution aims to bring robots closer to invertebrate animals and soft appendices of vertebrate animals (e.g., an elephant's trunk, a monkey's tail). This work aims to introduce the control theorist perspective to this novel development in robotics. We aim to remove the barriers to entry into this field by presenting existing results and future challenges using a unified language and within a coherent framework. Indeed, the main difficulty in entering this field is the wide variability of terminology and scientific backgrounds, making it quite hard to acquire a comprehensive view on the topic. Another limiting factor is that it is not obvious where to draw a clear line between the limitations imposed by the technology not being mature yet and the challenges intrinsic to this class of robots. In this work, we argue that the intrinsic effects are the continuum or multi-body dynamics, the presence of a non-negligible elastic potential field, and the variability in sensing and actuation strategies.

Abhishek Tomy, Matteo Razzanelli, Francesco Di Lauro et al.

When an epidemic spreads into a population, it is often unpractical or impossible to have a continuous monitoring of all subjects involved. As an alternative, algorithmic solutions can be used to infer the state of the whole population from a limited amount of measures. We analyze the capability of deep neural networks to solve this challenging task. Our proposed architecture is based on Graph Convolutional Neural Networks. As such it can reason on the effect of the underlying social network structure, which is recognized as the main component in the spreading of an epidemic. We test the proposed architecture with two scenarios modeled on the CoVid-19 pandemic: a generic homogeneous population, and a toy model of Boston metropolitan area.

Cosimo Della Santina, Dominic Lakatos, Antonio Bicchi et al.

Inspired by the vertebrate branch of the animal kingdom, articulated soft robots are robotic systems embedding elastic elements into a classic rigid (skeleton-like) structure. Leveraging on their bodies elasticity, soft robots promise to push their limits far beyond the barriers that affect their rigid counterparts. However, existing control strategies aiming at achieving this goal are either tailored on specific examples, or rely on model cancellations -- thus defeating the purpose of introducing elasticity in the first place. In a series of recent works, we proposed to implement efficient oscillatory motions in robots subject to a potential field, aimed at solving these issues. A main component of this theory are Eigenmanifolds, that we defined as nonlinear continuations of the classic linear eigenspaces. When the soft robot is initialized on one of these manifolds, it evolves autonomously while presenting regular -- and thus practically useful -- evolutions, called normal modes. In addition to that, we proposed a control strategy making modal manifolds attractors for the system, and acting on the total energy of the soft robot to move from a modal evolution to the other. In this way, a large class of autonomous behaviors can be excited, which are direct expression of the embodied intelligence of the soft robot. Despite the fact that the idea behind our work comes from physical intuition and preliminary experimental validations, the formulation that we have provided so far is however rather theoretical, and very much in need of an experimental validation. The aim of this paper is to provide such an experimental validation using as testbed the articulated soft leg. We will introduce a simplified control strategy, and we will test its effectiveness on this system, to implement swing-like oscillations. We plan to extend this validation with a soft quadruped.

Ivan Herreros, Xerxes D. Arsiwalla, Cosimo Della Santina et al.

How does our nervous system successfully acquire feedback control strategies in spite of a wide spectrum of response dynamics from different musculo-skeletal systems? The cerebellum is a crucial brain structure in enabling precise motor control in animals. Recent advances suggest that synaptic plasticity of cerebellar Purkinje cells involves molecular mechanisms that mimic the dynamics of the efferent motor system that they control allowing them to match the timing of their learning rule to behavior. Counter-Factual Predictive Control (CFPC) is a cerebellum-based feed-forward control scheme that exploits that principle for acquiring anticipatory actions. CFPC extends the classical Widrow-Hoff/Least Mean Squares by inserting a forward model of the downstream closed-loop system in its learning rule. Here we apply that same insight to the problem of learning the gains of a feedback controller. To that end, we frame a Model-Reference Adaptive Control (MRAC) problem and derive an adaptive control scheme treating the gains of a feedback controller as if they were the weights of an adaptive linear unit. Our results demonstrate that rather than being exclusively confined to cerebellar learning, the approach of controlling plasticity with a forward model of the subsystem controlled, an approach that we term as Model-Enhanced Least Mean Squares (ME-LMS), can provide a solution to wide set of adaptive control problems.