Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Alberto Dionigi has a total of 20 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • One to Rule Them All: Reinforcement Learning Policy Selection With Closed-Loop Sensitivity

    by Andrea Pupa, Alberto Dionigi, Paolo Robuffo Giordano, Cristian Secchi, Gabriele Costante

    In reinforcement learning-based control, reliable zero-shot transfer from simulation to real hardware remains a major open challenge. The offline training phase typically produces multiple models, each corresponding to a different policy, and choosing which model to deploy on physical hardware is a critical decision. However, there is no established methodology for this selection process. A common practice is to deploy the policy that achieves the highest reward in simulation, without evaluating its robustness to uncertainties. As a result, due to the sim-to-real gap, the selected policy may perform poorly when transferred to a physical robotic system, potentially leading to unsafe or undesirable behavior. This work proposes to exploit the notion of “closed-loop sensitivity” for applying it, for the first time, to the analysis and comparison of different reinforcement learning models, i.e., different policies, and to estimate their expected performance on real hardware. The proposed metric captures how the closed-loop system reacts to modeling errors, providing a quantitative measure of policy robustness. Both simulation results and real world experimental campaigns demonstrate that the proposed approach enables reliable zero-shot transfer to a real robot, while effectively handling uncertainties and reducing the risk of unsafe behaviors.

  • Open Access

    Towards trajectory following and vision-based collision avoidance for Micro Aerial Vehicles with Deep Reinforcement Learning

    by Raffaele Brilli, Alberto Dionigi, Marco Legittimo, Francesco Crocetti, Mirko Leomanni, Mario Luca Fravolini, Gabriele Costante


  • Open Access

    Comparison of DSO and ORB‐SLAM3 in Low‐Light Environments With Auxiliary Lighting and Deep Learning Based Image Enhancing

    by Francesco Crocetti, Raffaele Brilli, Alberto Dionigi, Mario L. Fravolini, Gabriele Costante, Paolo Valigi

    In the evolving landscape of robotic navigation, the demand for solutions capable of operating in challenging scenarios, such as low‐light environments, is increasing. This study investigates the performance of two state‐of‐the‐art (SOTA) visual simultaneous localization and mapping (VSLAM) algorithms, direct sparse odometry (DSO) and ORBSLAM3, in their monocular implementation, in the dark indoor scenarios where the only light source is provided by an auxiliary light system installed on a robot. A modified Pioneer3‐DX robot, equipped with a monocular camera, LED bars, and a lux meter, is utilized to collect a novel data set, “LUCID—Lighting Up Campus Indoor Spaces Data Set,” in real‐world, low‐light indoor environments. The data set includes image sequences enhanced using a generative adversarial network (GAN) to simulate varying levels of image enhancement. Through comprehensive experiments, we assess the performances of the V‐SLAM algorithm, considering the critical balance between maintaining adequate auxiliary illumination and enhancing. This study provides insights into the optimization of robotic navigation in lowlight conditions, paving the way for more robust and reliable autonomous navigation systems.


  • Quadrotor Control System Design for Robust Monocular Visual Tracking

    by Mirko Leomanni, Francesco Ferrante, Alberto Dionigi, Gabriele Costante, Paolo Valigi, Mario Luca Fravolini

    In this article, we present a control design framework for autonomous visual tracking of a dynamic target by a quadrotor vehicle. The considered problem poses important challenges related to the scale ambiguity, which affects monocular vision systems, and the lack of knowledge of the target dynamic response. These are addressed by leveraging a modular architecture that combines a state-of-the-art visual object tracker with a robust output feedback controller. In particular, by relying on the $\mu $ -synthesis technique, we design a controller providing suitable stability and performance guarantees. As opposed to previous solutions found in the literature, the proposed design does not require information about the target object depth or its geometry. The effectiveness of our approach is demonstrated through photorealistic simulations, showing that monocular tracking of aerial targets is consistently achieved.


  • Open Access

    D-VAT: End-to-End Visual Active Tracking for Micro Aerial Vehicles

    by Alberto Dionigi, Simone Felicioni, Mirko Leomanni, Gabriele Costante

    Visual active tracking is a growing research topic in robotics due to its key role in applications such as human assistance, disaster recovery, and surveillance. In contrast to passive tracking, active tracking approaches combine vision and control capabilities to detect and actively track the target. Most of the work in this area focuses on ground robots, while the very few contributions on aerial platforms still pose important design constraints that limit their applicability. To overcome these limitations, in this letter we propose D-VAT, a novel end-to-end visual active tracking methodology based on deep reinforcement learning that is tailored to micro aerial vehicle platforms. The D-VAT agent computes the vehicle thrust and angular velocity commands needed to track the target by directly processing monocular camera measurements. We show that the proposed approach allows for precise and collision-free tracking operations, outperforming different state-of-the-art baselines on simulated environments which differ significantly from those encountered during training. Moreover, we demonstrate a smooth real-world transition to a quadrotor platform with mixed-reality.


  • A Convex Programming Approach to Multipoint Optimal Motion Planning for Unicycle Robots

    by Mirko Leomanni, Giuseppe Mollica, Alberto Dionigi, Paolo Valigi, Gabriele Costante

    An important extension of point-to-point motion planning deals with navigating a robot optimally through a sequence of waypoints. The resulting multipoint problem is typically nonconvex due to the requirement to optimize the waypoint passage time, which has hindered the determination of globally optimal solutions. In this paper, a new motion planning technique is presented for multipoint problems involving unicycle robots. Our main finding is that, by adopting a cost function that weights a combination of path length and maneuver time, together with a suitable tightening of the control input constraints, such type of problems can be cast as a convex program. This can be solved in polynomial time to obtain a global optimum. The proposed motion planner is compared to time-optimal trajectory generation scheme based on nonlinear programming, and it can find better trajectories than this method while being computationally faster.


  • Open Access

    Integrating Sparse Learning-Based Feature Detectors into Simultaneous Localization and Mapping—A Benchmark Study

    by Giuseppe Mollica, Marco Legittimo, Alberto Dionigi, Gabriele Costante, Paolo Valigi

    Simultaneous localization and mapping (SLAM) is one of the cornerstones of autonomous navigation systems in robotics and the automotive industry. Visual SLAM (V-SLAM), which relies on image features, such as keypoints and descriptors to estimate the pose transformation between consecutive frames, is a highly efficient and effective approach for gathering environmental information. With the rise of representation learning, feature detectors based on deep neural networks (DNNs) have emerged as an alternative to handcrafted solutions. This work examines the integration of sparse learned features into a state-of-the-art SLAM framework and benchmarks handcrafted and learning-based approaches by comparing the two methods through in-depth experiments. Specifically, we replace the ORB detector and BRIEF descriptor of the ORBSLAM3 pipeline with those provided by Superpoint, a DNN model that jointly computes keypoints and descriptors. Experiments on three publicly available datasets from different application domains were conducted to evaluate the pose estimation performance and resource usage of both solutions.


  • E-VAT: An Asymmetric End-to-End Approach to Visual Active Exploration and Tracking

    by Alberto Dionigi, Alessandro Devo, Leonardo Guiducci, Gabriele Costante

    The development of visual tracking systems is becoming a major goal for the Robotics community. Most of the works dealing with this topic focus exclusively on passive tracking, where the target is confined within the camera’s field of view. Only a minority propose active approaches, capable not only of identifying the object to be tracked but also of producing motion control actions to maintain visual contact with it. However, all the methods introduced so far assume that the target is initially in the immediate proximity of the tracker. This represents an undesirable constraint on the applicability of these techniques, and it is to overcome this limitation that we propose a novel End-to-End Deep Reinforcement Learning based system, capable of both exploring the surrounding environment to find the target and then of tracking it. To do this, we develop a network consisting of two sub-components: i) the Target-Detection Network , which detects the target in the camera’s field-of-view , and ii) the Exploration and Tracking Network , which employs this information to switch between the exploration policy and the tracking policy with the goal of exploring the environment, finding the target and finally tracking it. Through different experiments, we demonstrate the effectiveness of our approach and its superior performance with respect to current state-of-the-art (SotA) methods.