Development of a physical mobile robot model in the Gazebo simulator

Svrha robotske neuronavigacije (u daljnjem tekstu RONNA) istraživanje je i razvitak inovativnih i konkurentnih robotskih sustava za neurokirurške primjene. Roboti u upotrebi su teški i stoga nisu idealni za preciznu navigaciju. Glavni cilj ovog rada je uspostaviti okvir koji ce omogućiti nastavak istraživanja mobilnosti RONNA-e. U sklopu ovog rada razvijena je i implementirana platforma za autonomnu navigaciju mobilnog robota s diferencijalnim pogonom. Implementacija algoritama za lokalizaciju te za globalno i lokalno planiranje putanje izvršena je pomoću Robotskog Operativnog Sustava (ROS). Kao dio rada napravljena je i analiza te verikacija dva različita algoritma za lokalno planiranje putanje. Izvedba algoritama DWA (eng. Dynamic Window Approach) i TEB (eng. Timed Elastic Band) procijenjena je kroz niz različitih konguracija u okviru tlocrta Regionalnog centra izvrsnosti za robotske tehnologije (CRTA) Sveučilista u Zagrebu. Uz ove procjene, razvijena su i dva vanjska programa koji prikupljaju i uspoređuju podatke o ubrzanjima robota. Algoritmi su morali zadovoljiti uvjete sigurnosti, reaktivnosti te optimalnosti putanje. U ovim testovima TEB algoritam ocijenjen je kao najprikladniji za potrebe ovog specifičnog robotskog sustava za neurokirurške primjene te za postojeći hardver.Robotic Neuronavigation's (hereinafter RONNA) aim is to explore and develop an innovative and competitive robotic system for neurosurgical applications. The robots in use are heavy and therefore not ideally suited for precise navigation. The main goal of this thesis is to establish a framework that will enable the continuation of RONNA mobility research. Within this thesis, a platform for autonomous navigation of a mobile robot with a dierential drive was developed and implemented. The implementation of algorithms for localization and for global and local path planning was performed using the Robot Operating System (ROS). As a part of this thesis, the verication of two dierent algorithms for local path planning was made and analyzed. DWA (Dynamic Window Approach) and TEB (Timed Elastic Band) algorithms performance was evaluated through a number of dierent configurations within the layout of the Regional Centre of Excellence for Robotics at the University of Zagreb. In addition to these evaluations, two external programs have also been developed that collect and compare acceleration data. Path planning algorithms had to meet the criteria of safety, reactivity and trajectory optimality. In this empiric test, the TEB algorithm was evaluated as the best t for the needs of this specic robotic system for neurosurgical applications and for the existing hardware