Human detection and tracking for safe motion in robot workspace

U ovom radu opisana je izrada sustava za detekciju i praćenje ljudi u svrhu sigurnog kretanja u okolini robotskog manipulatora. Realizirani sustav zasnovan je na ROS-u. Za prikupljanje podataka iz okoline koristi se Kinect RGB-D senzor. Korištenjem SPENCER razvojnog okvira, implementirana je metoda detekcije i praćenja ljudi koja računa opisne informacije o promatranim ljudima na temelju podataka snimljenih Kinect kamerom. Definirana su pravila za ograničenje kretanja robotskog manipulatora koji obavlja radnje u prostoru dijeljenom s ljudima. Informacije dobivene metodom detekcije i praćenja ljudi koriste se za provjeru uvjeta koji određuju ograničenja postavljena na kretanje robota. Ostvareno je upravljanje ABB IRB 2400L robotskim manipulatorom povezivanjem osobnog računala s IRC5 kontrolerom robota. Rad izrađenog sustava testiran je novim skupom podataka snimljenim Kinect kamerom.In this thesis, development of a human detection and tracking system for the purpose of safe movement in the robot environment is described. The developed system is based on ROS. Environment data is gathered using a Kinect RGB-D sensor. Using the SPENCER framework, a human detection and tracking method was implemented, which calculates descriptive information about observed humans using the data recorded with the Kinect camera. Rules for movement restriction of a robotic manipulator, operating in a workspace shared with humans, were defined. Conditions, that define the restrictions imposed to the manipulator movement, are checked using data from the human detection and tracking method. Control of the ABB IRB 2400L robotic manipulator was realized using a connection between a personal computer and an IRC5 robot controller. The developed system is tested using a new dataset recorded with a Kinect senso