Contributions to the performance analysis of continuum parallel robots

Continuum parallel robots (CPRs) are manipulators employing multiple flexible beams arranged in parallel and connected to a rigid end-effector. CPRs promise higher payload and accuracy than serial CRs while keeping great flexibility. As the risk of injury during accidental contacts between a human and a CPR should be reduced, CPRs may be used in large-scale collaborative tasks or assisted robotic surgery. There exist various CPR designs, but the prototype conception is rarely based on performance considerations, and the CPRs realization in mainly based on intuitions or rigid-link parallel manipulators architectures. This thesis focuses on the performance analysis of CPRs, and the tools needed for such evaluation, such as workspace computation algorithms. In particular, workspace computation strategies for CPRs are essential for the performance assessment, since the CPRs workspace may be used as a performance index or it can serve for optimal-design tools. Two new workspace computation algorithms are proposed in this manuscript, the former focusing on the workspace volume computation and the certification of its numerical results, while the latter aims at computing the workspace boundary only. Due to the elastic nature of CPRs, a key performance indicator for these robots is the stability of their equilibrium configurations. This thesis proposes the experimental validation of the equilibrium stability assessment on a real prototype, demonstrating limitations of some commonly used assumptions. Additionally, a performance index measuring the distance to instability is originally proposed in this manuscript. Differently from the majority of the existing approaches, the clear advantage of the proposed index is a sound physical meaning; accordingly, the index can be used for a more straightforward performance quantification, and to derive robot specifications.

Development of gripping devices based on electroadhesive forces

In a context of technological innovation, the aim of this thesis is to develop a technology that has gained interest in both scientific and industrial realms. This technology serves as a viable alternative to outdated and energy-consuming industrial systems. Electro-adhesive devices (EADs) leverage electrostatic forces for grasping objects or adhering to surfaces. The advantage of employing electrostatics lies in its adaptability to various materials without compromising the structure or chemistry of the object or surface. These benefits have led the industry to explore this technology as a replacement for costly vacuum systems and suction cups currently used for handling most products. Furthermore, the broad applicability of this technology extends to extreme environments, such as space with ultra-high vacuum conditions. Unfortunately, research in this area has yet to yield practical results for industrially effective gripper prototyping. This is primarily due to the inherent complexity of electro-adhesive technology, which operates on basic capacitive principles that does not find satisfying physical descriptions. This thesis aims to address these challenges through a series of studies, starting with the manufacturing process and testing of an EAD that has become the standard in our laboratory. It then delves into material and electrode geometry studies to enhance system performance, ultimately presenting potential industrial applications of the technology. All the presented results are encouraging, as they have yielded shear force values three times higher than those previously reported in the literature. The various applications have demonstrated the significant effectiveness of EADs as brakes or, more broadly, in exerting shear forces. This opens up the possibility of utilizing cutting-edge technologies to push the boundaries of technology to the fullest.

Analisi elettromiografica dei muscoli di arto superiore durante esercizi di reaching

Il presente lavoro di tesi si basa sull’analisi di segnali elettromiografici bipolari superficiali (sEMG) riguardanti movimenti di grasping e di reaching. L’elaborato fa parte del progetto Neurograsp, il cui obiettivo è correlare l’attività muscolare con le zone di attivazione corticale durante movimenti di arto superiore eseguiti da soggetti sani. I soggetti partecipanti eseguono dei movimenti per raggiungere una tra 5 posizioni target differenti, equidistanziate lungo un semicerchio posto davanti al soggetto, e successivamente tornare alla posizione iniziale. Ogni movimento viene ripetuto 10 volte, il target viene deciso in maniera random all’interno dell’acquisizione e ogni acquisizione è effettuata 6 volte. Ogni posizione è identificata da un led e la loro accensione permette di comunicare al soggetto quale posizione raggiungere e quando iniziare il movimento. I segnali di sEMG bipolare sono stati rilevati sui seguenti muscoli dell’arto superiore dominante: bicipite brachiale, tricipite brachiale, estensore del polso e flessore del polso. Queste informazioni sono state utilizzate per analizzare le attività muscolari durante i movimenti, nello specifico per identificare un apprendimento motorio da parte del singolo soggetto. In particolare, si vuole mettere a punto un protocollo di elaborazione dati al fine di valutare se c’è stato un effettivo miglioramento dei movimenti avanzando nelle acquisizioni e se è possibile individuare un pattern comune nell’esecuzione del movimento a una direzione specifica. Nel contesto Neurograsp, le informazioni ricavate da questo studio di tesi, forniscono indicazioni sull’utilità di andare ad individuare risposte di aree corticali relative all’apprendimento muscolare.

Mitigating non-Lambertian surfaces issues in Stereo Matching with Neural Radiance Fields

Depth estimation from images has long been regarded as a preferable alternative compared to expensive and intrusive active sensors, such as LiDAR and ToF. The topic has attracted the attention of an increasingly wide audience thanks to the great amount of application domains, such as autonomous driving, robotic navigation and 3D reconstruction. Among the various techniques employed for depth estimation, stereo matching is one of the most widespread, owing to its robustness, speed and simplicity in setup. Recent developments has been aided by the abundance of annotated stereo images, which granted to deep learning the opportunity to thrive in a research area where deep networks can reach state-of-the-art sub-pixel precision in most cases. Despite the recent findings, stereo matching still begets many open challenges, two among them being finding pixel correspondences in presence of objects that exhibits a non-Lambertian behaviour and processing high-resolution images. Recently, a novel dataset named Booster, which contains high-resolution stereo pairs featuring a large collection of labeled non-Lambertian objects, has been released. The work shown that training state-of-the-art deep neural network on such data improves the generalization capabilities of these networks also in presence of non-Lambertian surfaces. Regardless being a further step to tackle the aforementioned challenge, Booster includes a rather small number of annotated images, and thus cannot satisfy the intensive training requirements of deep learning. This thesis work aims to investigate novel view synthesis techniques to augment the Booster dataset, with ultimate goal of improving stereo matching reliability in presence of high-resolution images that displays non-Lambertian surfaces.

Axis control optimization: autotuning techniques and frequency response analysis of industrial mechatronic systems

In the field of industrial automation, there is an increasing need to use optimal control systems that have low tracking errors and low power and energy consumption. The motors we are dealing with are mainly Permanent Magnet Synchronous Motors (PMSMs), controlled by 3 different types of controllers: a position controller, a speed controller, and a current controller. In this thesis, therefore, we are going to act on the gains of the first two controllers by going to find, through the TwinCAT 3 software, what might be the best set of parameters. To do this, starting with the default parameters recommended by TwinCAT, two main methods were used and then compared: the method of Ziegler and Nichols, which is a tabular method, and advanced tuning, an auto-tuning software method of TwinCAT. Therefore, in order to analyse which set of parameters was the best,several experiments were performed for each case, using the Motion Control Function Blocks. Moreover, some machines, such as large robotic arms, have vibration problems. To analyse them in detail, it was necessary to use the Bode Plot tool, which, through Bode plots, highlights in which frequencies there are resonance and anti-resonance peaks. This tool also makes it easier to figure out which and where to apply filters to improve control.

Design and specification of a modular automatic testing solution

Electric vehicles and electronic components inside the vehicle are becoming increasingly important. The software as well starts to have a significant impact on modern high-end cars therefore a careful validation process needs to be implemented with the aim of having a bug free product when it is released. The software complexity increases and thus also the testing phases is more demanding. Test can be troublesome and, in some cases, boring and easy. The intelligence can be moved in test definition and writing rather than on test execution. The aim of this document is to start the definition of an automatic modular testing system capable to execute test cycles on systems that interacts with the CAN networks and with DUT that can be touched with a robotic arm. The document defines a first version of the system, in particular the hardware interface part with the aim of taking logs and execute test in an automated fashion with the test engineer can have a higher focus on the test definition and analysis rather than execution.

Continual learning on nano drones

Miniaturized flying robotic platforms, called nano-drones, have the potential to revolutionize the autonomous robots industry sector thanks to their very small form factor. The nano-drones’ limited payload only allows for a sub-100mW microcontroller unit for the on-board computations. Therefore, traditional computer vision and control algorithms are too computationally expensive to be executed on board these palm-sized robots, and we are forced to rely on artificial intelligence to trade off accuracy in favor of lightweight pipelines for autonomous tasks. However, relying on deep learning exposes us to the problem of generalization since the deployment scenario of a convolutional neural network (CNN) is often composed by different visual cues and different features from those learned during training, leading to poor inference performances. Our objective is to develop and deploy and adaptation algorithm, based on the concept of latent replays, that would allow us to fine-tune a CNN to work in new and diverse deployment scenarios. To do so we start from an existing model for visual human pose estimation, called PULPFrontnet, which is used to identify the pose of a human subject in space through its 4 output variables, and we present the design of our novel adaptation algorithm, which features automatic data gathering and labeling and on-device deployment. We therefore showcase the ability of our algorithm to adapt PULP-Frontnet to new deployment scenarios, improving the R2 scores of the four network outputs, with respect to an unknown environment, from approximately [−0.2, 0.4, 0.0,−0.7] to [0.25, 0.45, 0.2, 0.1]. Finally we demonstrate how it is possible to fine-tune our neural network in real time (i.e., under 76 seconds), using the target parallel ultra-low power GAP 8 System-on-Chip on board the nano-drone, and we show how all adaptation operations can take place using less than 2mWh of energy, a small fraction of the available battery power.