Publications

Brain-machine interfaces (BMIs) are alternative communication channels that have allowed healthy and disabled people to control external devices from brain signals. In the last decades, the growing attention towards neurorobotics has led to the proliferation of several BMI-based systems for controlling different devices including telepresence robots, powered wheelchairs, robotic arms, and upper/lower-limb exoskeletons. Despite the potentialities of these systems, it has emerged the necessity to create new forms of interaction between the human and the robot in order to increase the granularity of the user’s commands which are, in turn, translated into specific robot’s actions. In this preliminary work, we present how artificial intelligence can be exploited to design and tune a model able to convert the user’s intention into continuous robot’s movements.

The collaborative transport of objects between humans and robots is one of the main areas of focus in physical Human-Robot Interaction (pHRI). Ensuring the operator’s safety and maintaining collision-free motion of the robot during transportation are crucial challenges in this context. Consider a collaborative co-manipulation scenario where the operator modifies the trajectory being executed by the robot. In such cases, the robot may deviate from its previously calculated path, potentially resulting in collisions. In this work, we propose a method to estimate the maximum collision-free volume around the path of the robot. This volume represents the permissible deviation introduced by the human worker while ensuring that no collisions occur. To evaluate the effectiveness of the proposed algorithm, we test it in a real industrial scenario.

Hand-eye calibration is the problem of estimating the spatial transformation between a reference frame, usually the base of a robot arm or its gripper, and the reference frame of one or multiple cameras. Generally, this calibration is solved as a non-linear optimization problem, what instead is rarely done is to exploit the underlying graph structure of the problem itself. Actually, the problem of hand-eye calibration can be seen as an instance of the Simultaneous Localization and Mapping (SLAM) problem. Inspired by this fact, in this work we present a pose-graph approach to the hand-eye calibration problem that extends a recent state-of-the-art solution in two different ways: i) by formulating the solution to eye-on-base setups with one camera; ii) by covering multi-camera robotic setups. The proposed approach has been validated in simulation against standard hand-eye calibration methods. Moreover, a real application is shown. In both scenarios, the proposed approach overcomes all alternative methods. We release with this paper an open-source implementation of our graph-based optimization framework for multi-camera setups.

Robot grasping has been widely studied in the last decade. Recently, Deep Learning made possible to achieve remarkable results in grasp pose estimation, using depth and RGB images. However, only few works consider the choice of the object to grasp. Moreover, they require a huge amount of data for generalizing to unseen object categories. For this reason, we introduce the Few-shot Semantic Grasping task where the objective is inferring a correct grasp given only five labelled images of a target unseen object. We propose a new deep learning architecture able to solve the aforementioned problem, leveraging on a Few-shot Semantic Segmentation module. We have evaluated the proposed model both in the Graspnet dataset and in a real scenario. In Graspnet, we achieve 40,95% accuracy in the Few-shot Semantic Grasping task, outperforming baseline approaches. In the real experiments, the results confirmed the generalization ability of the network.

This paper presents Hi-ROS (Human Interaction in ROS), an open source framework focused on real-time accurate assessment of human motion. The system offers a series of tools to track multiple people in real-time by exploiting a calibrated camera network. No assumptions are made about the typology or number of cameras, nor about the body pose estimation algorithm used to extract the 3D poses of the people in the scene. The tools provided by Hi-ROS include a Skeleton Tracker to ensure temporal consistency of the detected poses, a Skeleton Merger to fuse the tracks from multiple cameras, thus limiting flickering phenomena, a Skeleton Optimizer to ensure limb length consistency, and a Skeleton Filter to perform real-time smoothing of the detected joint trajectories. Accuracy, tracking robustness, and real-time performance of the proposed system were evaluated on a public dataset, containing both single-person and multi-person sequences with up to 4 people interacting. The results obtained using different subsets of the proposed tools show how the complete Hi-ROS pipeline provides accurate and reliable estimates also in challenging scenarios, with a reduction of the RMSE of up to 27% with respect to a pure tracking approach. This work aims to push forward the development of unobtrusive human–robot interaction applications, multi-person automated posture analyses, rehabilitation performance assessments, and any possible application enabled by real-time accurate assessment of human motion via markerless motion capture.

Background: Lymphedema is a debilitating illness caused by insufficient lymph drainage, which can have serious physical and psychological consequences. Although water-based exercise can be useful, at present, little evidence is available regarding the outcomes of aquatic treatment for patients with lymphedema. Therefore, the aim of the present scoping review was to evaluate, from reported studies, the effects of water-based exercise on pain, limb motor function, quality of life (QoL), and limb volume among patients affected by primary and secondary upper and lower limb lymphedema. Methods: We performed a scoping review to examine clinical studies and randomized controlled trials reported in English from 2000 to 2021 by screening the MEDLINE (PubMed) and PEDro databases. Results: The search produced a total of 88 studies. Eight randomized controlled trials and one clinical study of patients with primary or secondary lymphedema of upper or lower limbs who had undergone water-based treatment were included in the present study. Most trials had focused on breast cancer-related lymphedema. The shoulder range of flexion, external rotation, and abduction have been shown to improve after performing a water-based exercise protocol. Some evidence has also demonstrated that the lymphedematous limb strength can improve. Moreover, water-based exercise seemed to improve pain perception and QoL for patients with upper or lower limb lymphedema. In contrast, in the control groups, the QoL showed a tendency to worsen over time. Although some studies had not reported beneficial effects on the lymphedematous limb volume, most of the studies examined had reported a reduction in volume, especially in the short term. No adverse events were reported in the included studies. Conclusions: The findings from the present review have shown the potential for aquatic exercise in lymphedema management. However, at the same time, the findings underline the multiple limitations resulting from the heterogeneity in the study populations and related physical activity protocols. The role of aquatic exercise in the conservative treatment of lymphedema requires further investigation in the future to define specific protocols of application.

A key building block for creating a circular economy is the ability to efficiently recover waste. For recycling to be profitable the purity of the separated fractions must be very high. The aim of the project is to implement a robotised waste sorting system to complement the current commercial solutions. The objective is to improve the quality and quantity of material recovered while limiting costs and labour use. This can be achieved thanks to advanced computer vision and robot manipulation techniques. The system will consist of two main components: (i) a vision system based on Deep Learning (DL) that combines several cameras to achieve high accuracy in material recognition; (ii) a manipulator robot that will sort objects based on feedback from the vision system. Grasp planning will exploit Reinforcement Learning (RL) to learn how to handle complex situations such as singling objects from a stack or disordered flow. The goal of innovation is twofold: to develop Artificial Intelligence techniques to be able to use low-cost sensors and to make system training simple and flexible for high reconfigurability to different types of waste.

Powered lower limb exoskeletons (LLEs) are innovative wearable robots that allow independent walking in people with severe gait impairments. Despite the recent advancements, the use of this promising technology is still restricted to clinical settings; uptake in real-life conditions as a device to promote user independence is still lacking due to the difficulty of controlling these devices in unstructured and complex environments. In this work, we propose a vision-assisted method for low obstacle avoidance to enhance the autonomy of LLEs. The exoskeleton collects information from the surroundings through a RGB-D camera to recognize and segment objects on the ground that might affect the walking pattern. Then, the method identifies suitable foothold positions. In addition, a novel iterative gait trajectory generator is proposed to automatically compute collision-free walking paths. We believe that re-thinking exoskeletons as semi-autonomous agents will represent not only the cornerstone to promote a more symbiotic human-exoskeleton interaction but may also pave the way for the use of this technology in the everyday life.

Self-driving vehicles and autonomous ground robots require a reliable and accurate method to analyze the traversability of the surrounding environment for safe navigation. This paper proposes and evaluates a real-time machine learning-based traversability analysis method that combines geometric features with a pyramid-polar space representation based on SVM classifiers. In particular, we show that by fusing geometric features with information stemming from coarser pyramid levels that account for a broader space portion, as well as integrating important implementation details, allows for a noticeable boost in performance and reliability. The main goal of this work is to demonstrate that traversability analysis is possible with effective results and in real-time even on cheaper hardware than expensive GPUs, e.g. CPU-only PCs. The proposed approach has been compared with state-of-the-art deep learning approaches on publicly available datasets of outdoor driving scenarios, running such algorithms both on GPU and CPU to compare runtimes. Our method can be fully executed on CPU and achieves results close to the best-in-class methods, runs faster, and requires fewer and less expensive hardware resources, consuming less than 30% electrical power with respect to deep learning models on embedded processing units. We release with this paper the open-source implementation of our method.

Neurorehabilitation with robotic devices requires a paradigm shift to enhance human-robot interaction. The coupling of robot assisted gait training (RAGT) with a brain-machine interface (BMI) represents an important step in this direction but requires better elucidation of the effect of RAGT on the user’s neural modulation. Here, we investigated how different exoskeleton walking modes modify brain and muscular activity during exoskeleton assisted gait. We recorded electroencephalographic (EEG) and electromyographic (EMG) activity from ten healthy volunteers walking with an exoskeleton with three modes of user assistance (i.e., transparent, adaptive and full assistance) and during free overground gait. Results identified that exoskeleton walking (irrespective of the exoskeleton mode) induces a stronger modulation of central mid-line mu (8-13 Hz) and low-beta (14-20 Hz) rhythms compared to free overground walking. These modifications are accompanied by a significant re-organization of the EMG patterns in exoskeleton walking. On the other hand, we observed no significant differences in neural activity during exoskeleton walking with the different assistance levels. We subsequently implemented four gait classifiers based on deep neural networks trained on the EEG data during the different walking conditions. Our hypothesis was that exoskeleton modes could impact the creation of a BMI-driven RAGT. We demonstrated that all classifiers achieved an average accuracy of 84.13 ± 3.49% in classifying swing and stance phases on their respective datasets. In addition, we demonstrated that the classifier trained on the transparent mode exoskeleton data can classify gait phases during adaptive and full modes with an accuracy of 78.3 ± 4.8%, while the classifier trained on free overground walking data fails to classify the gait during exoskeleton walking (accuracy of 59.4 ± 11.8%). These findings provide important insights into the effect of robotic training on neural activity and contribute to the advancement of BMI technology for improving robotic gait rehabilitation therapy.