Collaborative robotics is an increasing trend in the automation industry and is expected to keep increasing in the following years. Nonetheless, Cobots struggle to be a key factor in automating Small Medium Enterprises (SMEs). Indeed, their employment is limited to mere safety, while fundamental features, like programming intuitiveness and flexibility, remain untouched. However, if combined with AI-powered algorithms and the dexterity of the human operator, they can achieve high levels of flexibility and autonomy. To enhance and ease the introduction of automation inside SMEs, this work focuses on three main pillars: programming intuitiveness, robot autonomy to deal with unstructured environments and Task modifications, and human-robot interactions. This thesis aims to introduce innovative paradigms that account for the presence of unskilled operators in the production line, the production process changes caused by highly personalised products, and human-robot collaboration. In particular, Programming by Demonstration has been leveraged as an intuitive means to transfer Task knowledge from human to robot. From the demonstrations, which can be performed by the human through the robot (Kinesthetic Teaching) or directly by the human inside a controlled environment (Virtual Reality), the robot creates a database of Skills that can be exploited to achieve complex Tasks. The robot leverages the Skills database to achieve autonomy in terms of Task modifications and unexpected events. The demonstrations provide each Skills with a semantic description that planners can leverage to issue the robotic agent with the most appropriate action. The Skills are executed by the robot accordingly with the motion primitives performed during the demonstrations while controlling that the workspace gets modified as expected. On the other hand, this thesis aims also at modelling human-robot interactions that cannot be inferred from demonstrations. Therefore, it is necessary to introduce more rigid but always intuitive paradigms that grant seamless human-robot collaboration. The operator is provided with an intuitive web-based interface to compose the human-robot workflow in a PLC logic fashion. In particular, PLC logic is combined with tools semantics to recognise the operator’s actions, notifying him/her in case of problems. Indeed, such architecture grants human supervision and synchronisation with the robot during the Task, ensuring high-quality control and production efficiency. The effectiveness of the proposed paradigms has been validated in realistic industrial-like scenarios involving several volunteers with different robotic backgrounds.
La robotica collaborativa e’ un trend in aumento nell’industria dell’automazione e ci si aspetta che continui a crescere negli anni a venire. I Cobot pero’ fanno ancora fatica ad imporsi come un fattore fondamentale nell’automazione delle piccole medie imprese (PMI) dato che, fino ad adesso, il loro utilizzo e’ limitato alla sola sicurezza. Tale utilizzo non sfrutta le loro vere potenzialita’, come l’intuitivita’ di programmazione e flessibilita’. Inoltre, se si combinanono con algoritmi basati sull’AI e la manualita’ di un operatore umano, possono raggiungere alti livelli di flessibilita’ ed autonomia. Dunque, per migliorare e facilitare l’introduzione dell’automazione dentro le PMI, questo lavoro si concentra su tre pilastri fondamentali: intuitivita’ di programmazione, autonomia del robot nel gestire scenari complessi e cambi di produzione e permettere l’interazione uomo-robot. Questa tesi mira a introdurre paradigmi innovativi che tengono conto della presenza di operatori non esperti nella linea produttiva, della necessita’ di cambiare spesso il processo produttivo a causa di prodotti altamente personalizzati e permettere la collaborazione uomo-robot. Piu’ nel particolare, Programming by Demonstration e’ stata sfruttata come un mezzo intuitivo per trasferire conoscenza del Task dall’uomo all’operatore. Dalla dimostrazione, che puo’ essere fatta dall’umano attraverso il robot (insegnamento cinestetico) o direttamente dall’operatore dentro un ambiente controllato (realta’ virtuale), il robot crea un database di Skills che possono essere sfruttate per realizzare Task complessi. Il robot sfrutta tale database di Skills per raggiungere l’autonomia sia riguardo la modifica del Task ma anche per eventuali eventi inaspettati. Difatti, ogni Skills e’ stata descritta semanticamente e puo’ essere programmata da dei pianificatori e mandata al robot. Il robot dunque esegue la Skill con le primitive di moto apprese durante la dimostrazione mentre verifica che l’ambiente si modifichi come si aspetta. Dall’altra parte, le interazioni uomo-robot non possono essere trasferite tramite una dimostrazione, quindi e’ necessario introdurre un paradigma piu’ rigido, ma sempre intuitivo, che permette la collaborazione uomo robot. Si e’ dunque fornita all’operatore un’ interfaccia web che gli permette di modellare tramite una logica PLC il flusso di lavoro. Tale logica e’ combinata con la semantica degli oggetti usati durante il Task al fine di riconoscere le azioni dell’operatore, notificandolo in caso di errore o problemi. Difatti, tale architettura, permette la supervisione e la sincronizzazione dell’operatore col robot durante il Task, assicurando un alto controllo qualita’ e efficienza nella produzione. L’efficacia dei paradigmi proposti e’ stata validata tramite applicazioni in scenari industriali che hanno coinvolto vari volontari con diversi background robotici.
Enhancing flexibility, autonomy and interaction in collaborative robotics
LUCCI, NICCOLÒ
2023/2024
Abstract
Collaborative robotics is an increasing trend in the automation industry and is expected to keep increasing in the following years. Nonetheless, Cobots struggle to be a key factor in automating Small Medium Enterprises (SMEs). Indeed, their employment is limited to mere safety, while fundamental features, like programming intuitiveness and flexibility, remain untouched. However, if combined with AI-powered algorithms and the dexterity of the human operator, they can achieve high levels of flexibility and autonomy. To enhance and ease the introduction of automation inside SMEs, this work focuses on three main pillars: programming intuitiveness, robot autonomy to deal with unstructured environments and Task modifications, and human-robot interactions. This thesis aims to introduce innovative paradigms that account for the presence of unskilled operators in the production line, the production process changes caused by highly personalised products, and human-robot collaboration. In particular, Programming by Demonstration has been leveraged as an intuitive means to transfer Task knowledge from human to robot. From the demonstrations, which can be performed by the human through the robot (Kinesthetic Teaching) or directly by the human inside a controlled environment (Virtual Reality), the robot creates a database of Skills that can be exploited to achieve complex Tasks. The robot leverages the Skills database to achieve autonomy in terms of Task modifications and unexpected events. The demonstrations provide each Skills with a semantic description that planners can leverage to issue the robotic agent with the most appropriate action. The Skills are executed by the robot accordingly with the motion primitives performed during the demonstrations while controlling that the workspace gets modified as expected. On the other hand, this thesis aims also at modelling human-robot interactions that cannot be inferred from demonstrations. Therefore, it is necessary to introduce more rigid but always intuitive paradigms that grant seamless human-robot collaboration. The operator is provided with an intuitive web-based interface to compose the human-robot workflow in a PLC logic fashion. In particular, PLC logic is combined with tools semantics to recognise the operator’s actions, notifying him/her in case of problems. Indeed, such architecture grants human supervision and synchronisation with the robot during the Task, ensuring high-quality control and production efficiency. The effectiveness of the proposed paradigms has been validated in realistic industrial-like scenarios involving several volunteers with different robotic backgrounds.File | Dimensione | Formato | |
---|---|---|---|
PhD_Thesis_Lucci.pdf
accessibile in internet per tutti a partire dal 27/02/2025
Descrizione: Tesi
Dimensione
20.77 MB
Formato
Adobe PDF
|
20.77 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/217253