In the novel 5.0 industrial paradigm, humans and robots are considered as even collaborators. As such, they equally partake in the workflow, sharing spaces, tasks and information with each other. Owing to the latest advancements in the field, humans and robots are now allowed to physically interact with each other, either intentionally, for collaboratively performing dexterous tasks, or accidentally, in case of unforeseen collisions during task execution. In this context, a fruitful collaboration calls for the robotic application to comply with numerous safety and efficiency requirements. First and foremost, the safety of the human operator has to be ensured at all times. This requires that the behavior implemented by the robot is robustly stable during the interaction phase, as well as in free motion. At the same time, each indvidual task can present a different set of specific safety requirements, such as preserving a given separation distance between the human and the robot, avoiding joint limits or self collision. Furthermore, as human-robot shared workspaces are extremely dynamic environments, an opportune degree of flexibility is required from the robotic behavior, in order to adapt on the fly to the strongly time-varying conditions of the environment and of the different tasks at hand. We first address these issues simultaneously by incorporating this set of requirements into a single and generalizable control architecture for a flexible and effective human robot collaboration. During the course of this thesis, we showcase the capabilities of the architecture and how this can be deployed in various collaborative contexts. As energy-aware control has proven extremely successfull in modelling the interaction between physical systems, we exploit state of the art energy-based control strategies in order to ensur a robustly stable behavior of the robot. In particular, we leverage energy tanks in order to bound the energy injected in the controlled system. At the same time, we expand and build upon the traditional usage of energy tanks, in order to maximize their performarce in collaborative human-robot scenarios, as well as during the interaction with unknown / unmodeled environments.
Nel nuovo paradigma industriale 5.0, esseri umani e robot sono considerati alla pari dei collaboratori. In quanto tali, prendono ugualmente parte al flusso di lavoro, condividendo spazi, attività e informazioni tra loro. Grazie agli ultimi progressi nel campo, ora gli esseri umani e i robot possono interagire fisicamente tra loro, sia intenzionalmente, per eseguire in modo collaborativo compiti abili, sia accidentalmente, in caso di collisioni impreviste durante l'esecuzione delle attività. In questo contesto, una proficua collaborazione richiede che l'applicazione robotica soddisfi numerosi requisiti di sicurezza ed efficienza. In primo luogo, deve essere garantita in ogni momento la sicurezza dell'operatore umano. Ciò richiede che il comportamento implementato dal robot sia solidamente stabile durante la fase di interazione, così come in movimento libero. Allo stesso tempo, ogni singola attività può presentare un insieme diverso di requisiti di sicurezza specifici, come preservare una determinata distanza di separazione tra l'uomo e il robot, evitare limiti articolari o autocollisione. Inoltre, poiché gli spazi di lavoro condivisi uomo-robot sono ambienti estremamente dinamici, è richiesto un opportuno grado di flessibilità dal comportamento robotico, per adattarsi al volo alle condizioni fortemente variabili nel tempo dell'ambiente e dei diversi compiti a portata di mano. Per prima cosa affrontiamo questi problemi contemporaneamente incorporando questo insieme di requisiti in un'architettura di controllo unica e generalizzabile per una collaborazione flessibile ed efficace tra robot umani. Nel corso di questa tesi, mostriamo le capacità dell'architettura e come questa può essere implementata in vari contesti collaborativi. Poiché il controllo consapevole dell'energia si è dimostrato estremamente efficace nella modellazione dell'interazione tra i sistemi fisici, sfruttiamo strategie di controllo basate sull'energia all'avanguardia per garantire un comportamento robusto e stabile del robot. In particolare, sfruttiamo i serbatoi di energia per vincolare l'energia immessa nel sistema controllato. Allo stesso tempo, espandiamo e ci basiamo sull'uso tradizionale dei serbatoi di energia, al fine di massimizzare le loro prestazioni in scenari collaborativi uomo-robot, così come durante l'interazione con ambienti sconosciuti/non modellati.
Strategie di controllo avanzate basate sull'energia per un'interazione fisica uomo-robot flessibile e sicura
BENZI, FEDERICO
2023
Abstract
In the novel 5.0 industrial paradigm, humans and robots are considered as even collaborators. As such, they equally partake in the workflow, sharing spaces, tasks and information with each other. Owing to the latest advancements in the field, humans and robots are now allowed to physically interact with each other, either intentionally, for collaboratively performing dexterous tasks, or accidentally, in case of unforeseen collisions during task execution. In this context, a fruitful collaboration calls for the robotic application to comply with numerous safety and efficiency requirements. First and foremost, the safety of the human operator has to be ensured at all times. This requires that the behavior implemented by the robot is robustly stable during the interaction phase, as well as in free motion. At the same time, each indvidual task can present a different set of specific safety requirements, such as preserving a given separation distance between the human and the robot, avoiding joint limits or self collision. Furthermore, as human-robot shared workspaces are extremely dynamic environments, an opportune degree of flexibility is required from the robotic behavior, in order to adapt on the fly to the strongly time-varying conditions of the environment and of the different tasks at hand. We first address these issues simultaneously by incorporating this set of requirements into a single and generalizable control architecture for a flexible and effective human robot collaboration. During the course of this thesis, we showcase the capabilities of the architecture and how this can be deployed in various collaborative contexts. As energy-aware control has proven extremely successfull in modelling the interaction between physical systems, we exploit state of the art energy-based control strategies in order to ensur a robustly stable behavior of the robot. In particular, we leverage energy tanks in order to bound the energy injected in the controlled system. At the same time, we expand and build upon the traditional usage of energy tanks, in order to maximize their performarce in collaborative human-robot scenarios, as well as during the interaction with unknown / unmodeled environments.File | Dimensione | Formato | |
---|---|---|---|
Tesi Definitiva Benzi Federico.pdf
embargo fino al 22/03/2026
Dimensione
24.95 MB
Formato
Adobe PDF
|
24.95 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/79413
URN:NBN:IT:UNIMORE-79413