Collaborative Robotics is one of the fastest growing areas in the field of Industrial Robotics. Unlike traditional Robot a Collaborative Robot, or Cobot, can work at close contact with a human operator cooperating with him during the execution of a task. In such a scenario safety cages that divide the human workspace from the robot working area, become useless and can be removed: this leads to a great space saving and increase the productivity of the Robot because it can continue to work also if the human operator is in the same working area of the Robot. On one side the Human-Robot Interaction offers to companies, especially to the medium-small ones, new alternatives more flexible and efficient to the classical production lines but on the other side introduces new problematic linked with safety, since as already said now human operator and Robot work together and directly interact also when the robot is moving. This thesis work is dedicated to the study of the Safe Human-Robot Interaction, proposing to describe in an analytical way the problem of Safety during a human-Robot collaborative operation keeping the level of productivity as high as possible. It is here presented in detail a Strategy Control technique to guarantee the safety of the operator while he works and cooperate with the Robot. This Control Strategy will be analyzed and then improved trying to increase the level of productivity and to eliminate some issue linked with the current Control Strategy used. A vision sensor, an RGB-D camera, is employed by the Control Strategy in order to give the Robot a perception of the environment in which it operates and of the possible obstacle like a human operator. In this sense the Control Strategy integrates the system function that has to be already embedded on a Cobot and allow, sending to the Robot Controller Unit the information coming from the external sensor, to anticipate the activation of the emergency stop procedures that means a great loss of productivity. In the following the Control Strategy different levels are presented: 1. A speed Rescaling Algorithm computes a parameter that allows the Robot to reduce its velocity and to slow down if an operator is present in the vicinity of one or more part of the Robot and to stop it if the distance between the two agents becomes too small. Two possible approaches are presented: the first is already used and has already been tested in a real Human-Robot cooperative interaction whereas the second method aims to improve the results obtained with the implementation of the first which resulted to be too conservative, influencing the productivity of the Robot. 2. A Path Re-Planning Algorithm based on the generation of Waypoints through which the Robot can dodge an obstacle and continue its pre-assigned task without stop its motion if the obstacle comes too close to it. Also in this case two algorithm will be presented and analyzed: after the introduction of a first method, which turned out to be in some case not reliable in finding a valid dodging path, a second method will be described to overcome some problematics found during a dodge of a human operator.
La Robotica Collaborativa è uno dei settori più in crescita della robotica industriale. Al contrario dei tradizionali Robot industriali un Robot Collaborativo, o Cobot, può essere messo a lavorare a stretto contatto con un operatore umano interagendo con quest’ultimo durante lo svolgimento di un operazione. In uno scenario simile diventano quindi inutili le barriere protettive che separano lo spazio di lavoro dell’uomo da quello del Robot: ciò comporta un grande risparmio di spazio ma aumenta anche la produttività in quanto il Robot pu`o continuare a lavorare anche quando l’operatore è presente nella sua area di lavoro. Se da un lato l’interazione Uomo-Robot offre alle industrie, soprattutto a quelle medio-piccole, nuove alternative più flessibili e efficienti alle classiche linee di produzione dall’altro introduce nuove problematiche relative alla sicurezza, dato che uomo e Robot ora si trovano a lavorare a stretto contatto senza più essere separati da barriere fisiche. Questa tesi si occupa dello studio dell’Interazione Sicura Uomo-Robot, proponendosi di descrivere in modo analitico il problema della Sicurezza durante un’interazione collaborativa tra uomo e Robot mantenendo sempre più alto possibile il livello di produttività. E’ presentata quindi in dettaglio una Strategia di Controllo per garantire la sicurezza dell’operatore mentre si trova a collaborare con il Robot, che successivamente verrà poi integrata e migliorata per aumentarne il livello di produttività. La Strategia di Controllo fa uso di un sensore di visione, una telecamera RGB-D, per fornire al robot una percezione dell’ambiente di lavoro in cui lavora e della presenza di eventuali ostacoli come un operatore umano. Essa rappresenta quindi un’integrazione ai sistemi di sicurezza che devono già essere presenti su un Cobot e permette, comunicando al controllore del robot le informazioni provenienti dai sensori esterni, di anticipare l’attivazione di procedure di emergenza, per esempio dovute a delle collisioni, che richiederebbero l’intervento umano per ripristinare il sistema. Nel dettaglio la Strategia di Controllo è composta da due livelli: 1. Un Algoritmo di Riscalatura Cinematica della velocità calcola un parametro che permette al Robot di rallentare se un operatore umano si trova in vicinanza di una o più parti del robot stesso fino a fermarsi completamente se la distanza tra i due agenti è diventata troppo piccola. Due possibili approcci sono presentati: il primo è già stato utilizzato e testato in una situazione reale di interazione uomo-Robot mentre il secondo si propone di migliorare i risultati ottenuti con il primo che è risultato essere in alcune situazioni troppo conservativo, influendo così sulla produttività del Robot. 2. Una Strategia di Schivata basata sul calcolo di punti di via (Waypoint) attraverso i quali il robot può schivare un ostacolo e continuare l’operazione a lui assegnata senza fermarsi ogni volta che la distanza con l’ostacolo diventa troppo piccola. Anche in questo caso vengono presentati due algoritmi: dopo aver introdotto un primo metodo, che si è rivelato in alcuni casi inaffidabile, un secondo metodo viene presentato per superare le problematiche riscontrate durante la schivata di un operatore.
Control strategies for safe human-robot collaborative interaction
RAIMONDI, LORENZO;SABBADINI, MARCO
2017/2018
Abstract
Collaborative Robotics is one of the fastest growing areas in the field of Industrial Robotics. Unlike traditional Robot a Collaborative Robot, or Cobot, can work at close contact with a human operator cooperating with him during the execution of a task. In such a scenario safety cages that divide the human workspace from the robot working area, become useless and can be removed: this leads to a great space saving and increase the productivity of the Robot because it can continue to work also if the human operator is in the same working area of the Robot. On one side the Human-Robot Interaction offers to companies, especially to the medium-small ones, new alternatives more flexible and efficient to the classical production lines but on the other side introduces new problematic linked with safety, since as already said now human operator and Robot work together and directly interact also when the robot is moving. This thesis work is dedicated to the study of the Safe Human-Robot Interaction, proposing to describe in an analytical way the problem of Safety during a human-Robot collaborative operation keeping the level of productivity as high as possible. It is here presented in detail a Strategy Control technique to guarantee the safety of the operator while he works and cooperate with the Robot. This Control Strategy will be analyzed and then improved trying to increase the level of productivity and to eliminate some issue linked with the current Control Strategy used. A vision sensor, an RGB-D camera, is employed by the Control Strategy in order to give the Robot a perception of the environment in which it operates and of the possible obstacle like a human operator. In this sense the Control Strategy integrates the system function that has to be already embedded on a Cobot and allow, sending to the Robot Controller Unit the information coming from the external sensor, to anticipate the activation of the emergency stop procedures that means a great loss of productivity. In the following the Control Strategy different levels are presented: 1. A speed Rescaling Algorithm computes a parameter that allows the Robot to reduce its velocity and to slow down if an operator is present in the vicinity of one or more part of the Robot and to stop it if the distance between the two agents becomes too small. Two possible approaches are presented: the first is already used and has already been tested in a real Human-Robot cooperative interaction whereas the second method aims to improve the results obtained with the implementation of the first which resulted to be too conservative, influencing the productivity of the Robot. 2. A Path Re-Planning Algorithm based on the generation of Waypoints through which the Robot can dodge an obstacle and continue its pre-assigned task without stop its motion if the obstacle comes too close to it. Also in this case two algorithm will be presented and analyzed: after the introduction of a first method, which turned out to be in some case not reliable in finding a valid dodging path, a second method will be described to overcome some problematics found during a dodge of a human operator.File | Dimensione | Formato | |
---|---|---|---|
2018_12_Raimondi_Sabbadini.pdf
non accessibile
Descrizione: Testo della tesi
Dimensione
82.32 MB
Formato
Adobe PDF
|
82.32 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/10589/144921