Virtual sensing technology applied to a swarm of autonomous robots.

This thesis proposes a novel technology in the field of swarm robotics that allows a swarm of robots to sense a virtual environment through virtual sensors. Virtual sensing is a desirable and helpful technology in swarm robotics research activity, because it allows the researchers to efficiently and quickly perform experiments otherwise more expensive and time consuming, or even impossible. In particular, we envision two useful applications for virtual sensing technology. On the one hand, it is possible to prototype and foresee the effects of a new sensor on a robot swarm, before producing it. On the other hand, thanks to this technology it is possible to study the behaviour of robots operating in environments that are not easily reproducible inside a lab for safety reasons or just because physically infeasible. The use of virtual sensing technology for sensor prototyping aims to foresee the behaviour of the swarm enhanced with new or more powerful sensors, without producing the hardware. Sensor prototyping can be used to tune a new sensor or perform performance comparison tests between alternative types of sensors. This kind of prototyping experiments can be performed through the presented tool, that allows to rapidly develop and test software virtual sensors of different typologies and quality, emulating the behaviour of several hardware real sensors. By investigating on which sensors is better to invest, a researcher can minimize the sensors’ production cost while achieving a given swarm performance. Through augmented reality, it is possible to test the performance of the swarm in a desired virtual environment that cannot be set into the lab for physical, logistic or economical reasons. The virtual environment is sensed by the robots through properly designed virtual sensors. Virtual sensing technology allows a researcher to quickly carry out real robots experiment in challenging scenarios without all the required hardware and environment.

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.

Modelling and simulating in systems biology: an approach based on multi-agent systems

Systems Biology is an innovative way of doing biology recently raised in bio-informatics contexts, characterised by the study of biological systems as complex systems with a strong focus on the system level and on the interaction dimension. In other words, the objective is to understand biological systems as a whole, putting on the foreground not only the study of the individual parts as standalone parts, but also of their interaction and of the global properties that emerge at the system level by means of the interaction among the parts. This thesis focuses on the adoption of multi-agent systems (MAS) as a suitable paradigm for Systems Biology, for developing models and simulation of complex biological systems. Multi-agent system have been recently introduced in informatics context as a suitabe paradigm for modelling and engineering complex systems. Roughly speaking, a MAS can be conceived as a set of autonomous and interacting entities, called agents, situated in some kind of nvironment, where they fruitfully interact and coordinate so as to obtain a coherent global system behaviour. The claim of this work is that the general properties of MAS make them an effective approach for modelling and building simulations of complex biological systems, following the methodological principles identified by Systems Biology. In particular, the thesis focuses on cell populations as biological systems. In order to support the claim, the thesis introduces and describes (i) a MAS-based model conceived for modelling the dynamics of systems of cells interacting inside cell environment called niches. (ii) a computational tool, developed for implementing the models and executing the simulations. The tool is meant to work as a kind of virtual laboratory, on top of which kinds of virtual experiments can be performed, characterised by the definition and execution of specific models implemented as MASs, so as to support the validation, falsification and improvement of the models through the observation and analysis of the simulations. A hematopoietic stem cell system is taken as reference case study for formulating a specific model and executing virtual experiments.

Factors affecting the mission time of an autonomous mobile robot: a theoretical and experimental investigation

I vantaggi dell’Industria 4.0 hanno stravolto il manufacturing. Ma cosa vuol dire "Industria 4.0"? Essa è la nuova frontiera del manufacturing, basata su princìpi che seguono i passi avanti dei sistemi IT e della tecnologia. Dunque, i suoi pilastri sono: integrazione, verticale e orizzontale, digitalizzazione e automazione. L’Industria 4.0 coinvolge molte aree della supply chain, dai flussi informativi alla logistica. In essa e nell’intralogistica, la priorità è sviluppare dei sistemi di material handling flessibili, automatizzati e con alta prontezza di risposta. Il modello ideale è autonomo, in cui i veicoli fanno parte di una flotta le cui decisioni sono rese decentralizzate grazie all'alta connettività e alla loro abilità di collezionare dati e scambiarli rapidamente nel cloud aziendale.Tutto ciò non sarebbe raggiungibile se ci si affidasse a un comune sistema di trasporto AGV, troppo rigido e centralizzato. La tesi si focalizza su un tipo di material handlers più flessibile e intelligente: gli Autonomous Mobile Robots. Grazie alla loro intelligenza artificiale e alla digitalizzazione degli scambi di informazioni, interagiscono con l’ambiente per evitare ostacoli e calcolare il percorso ottimale. Gli scenari dell’ambiente lavorativo determinano perdite di tempo nel tragitto dei robot e sono queste che dovremo studiare. Nella tesi, i vantaggi apportati dagli AMR, come la loro decentralizzazione delle decisioni, saranno introdotti mediante una literature review e poi l’attenzione verterà sull’analisi di ogni scenario di lavoro. Fondamentali sono state le esperienze nel Logistics 4.0 Lab di NTNU, per ricreare fisicamente alcuni scenari. Inoltre, il software AnyLogic sarà usato per riprodurre e simulare tutti gli scenari rilevanti. I risultati delle simulazioni verranno infine usati per creare un modello che associ ad ogni scenario rilevante una perdita di tempo, attraverso una funzione. Per questo saranno usati software di data analysis come Minitab e MatLab.

Integration of ROS for the control of industrial robotic systems in a proprietary environment

The paper deals with the integration of ROS, in the proprietary environment of the Marchesini Group company, for the control of industrial robotic systems. The basic tools of this open-source software are deeply studied to model a full proprietary Pick and Place manipulator inside it, and to develop custom ROS nodes to calculate trajectories; speaking of which, the URDF format is the standard to represent robots in ROS and the motion planning framework MoveIt offers user-friendly high-level methods. The communication between ROS and the Marchesini control architecture is established using the OPC UA standard; the tasks computed are transmitted offline to the PLC, supervisor controller of the physical robot, because the performances of the protocol don’t allow any kind of active control by ROS. Once the data are completely stored at the Marchesini side, the industrial PC makes the real robot execute a trajectory computed by MoveIt, so that it replicates the behaviour of the simulated manipulator in Rviz. Multiple experiments are performed to evaluate in detail the potential of ROS in the planning of movements for the company proprietary robots. The project ends with a small study regarding the use of ROS as a simulation platform. First, it is necessary to understand how a robotic application of the company can be reproduced in the Gazebo real world simulator. Then, a ROS node extracts information and examines the simulated robot behaviour, through the subscription to specific topics.

Feedback control of underactuated cable-driven parallel robots

In this work an Underactuated Cable-Driven Parallel Robot (UACDPR) that operates in the three dimensional Euclidean space is considered. The End-Effector has 6 degrees of freedom and is actuated by 4 cables, therefore from a mechanical point of view the robot is defined underconstrained. However, considering only three controlled pose variables, the degree of redundancy for the control theory can be considered one. The aim of this thesis is to design a feedback controller for a point-to-point motion that satisfies the transient requirements, and is capable of reducing oscillations that derive from the reduced number of constraints. A force control is chosen for the positioning of the End-Effector, and error with respect to the reference is computed through data measure of several sensors (load cells, encoders and inclinometers) such as cable lengths, tension and orientation of the platform. In order to express the relation between pose and cable tension, the inverse model is derived from the kinematic and dynamic model of the parallel robot. The intrinsic non-linear nature of UACDPRs systems introduces an additional level of complexity in the development of the controller, as a result the control law is composed by a partial feedback linearization, and damping injection to reduce orientation instability. The fourth cable allows to satisfy a further tension distribution constraint, ensuring positive tension during all the instants of motion. Then simulations with different initial conditions are presented in order to optimize control parameters, and lastly an experimental validation of the model is carried out, the results are analysed and limits of the presented approach are defined.

Smart freight transport in urban areas: a crowd-shipping model using autonomous vehicles

L'avanzamento dell'e-commerce e l'aumento della densità abitativa nel centro città sono elementi che incentivano l'incremento della richiesta merci all'interno dei centri urbani. L'attenzione all'impatto ambientale derivante da queste attività operative è un punto focale oggetto di sempre maggiore interesse. Attraverso il seguente studio, l'obiettivo è definire attuali e potenziali soluzioni nell'ambito della logistica urbana, con particolare interesse alle consegne dell'ultimo miglio. Una soluzione proposta riguarda la possibilità di sfruttare la capacità disponibile nei flussi generati dalla folla per movimentare merce, pratica nota sotto il nome di Crowd-shipping. L'idea consiste nella saturazione di mezzi già presenti nella rete urbana al fine di ridurre il numero di veicoli commerciali e minimizzare le esternalità negative annesse. A supporto di questa iniziativa, nell'analisi verranno considerati veicoli autonomi elettrici a guida autonoma. La tesi è incentrata sulla definizione di un modello di ottimizzazione matematica, che mira a designare un network logistico-distributivo efficiente per le consegne dell'ultimo miglio e a minimizzare le distanze degli attori coinvolti. Il problema proposto rappresenta una variante del Vehicle Routing Problem con time windows e multi depots. Il problema è NP-hard, quindi computazionalmente complesso per cui sarà necessario, in fase di analisi, definire un approccio euristico che permetterà di ottenere una soluzione sub-ottima in un tempo di calcolo ragionevole per istanze maggiori. L'analisi è stata sviluppata nell'ambiente di sviluppo Eclipse, attraverso il risolutore Cplex, in linguaggio Java. Per poterne comprendere la validità, è prevista un'ultima fase in cui gli output del modello ottimo e dell'euristica vengono confrontati tra loro su parametri caratteristici. Bisogna tuttavia considerare che l' utilizzo di sistemi cyber-fisici a supporto della logistica non può prescindere da un costante sguardo verso il progresso.

Distributed collision-free traffic management via vertex cover for multi-vehicle systems

In this thesis, we state the collision avoidance problem as a vertex covering problem, then we consider a distributed framework in which a team of cooperating Unmanned Vehicles (UVs) aim to solve this optimization problem cooperatively to guarantee collision avoidance between group members. For this purpose, we implement a distributed control scheme based on a robust Set-Theoretic Model Predictive Control ( ST-MPC) strategy, where the problem involves vehicles with independent dynamics but with coupled constraints, to capture required cooperative behavior.