Development of Algorithm for Trajectory Planning with Obstacle Avoidance for Mobile Robot

Kroz ovaj diplomski rad opisana je izrada mobilnog robota koji primjenom umjetne inteligencije pronalazi optimalnu trajektoriju u svrhu dolaska na cilj, pritom izbjegavajući prepreke. Opisana je izrada komunikacije za povezivanje mobilnog robota sa ureĎajima na lokalnoj mreţi. Opisan je proces pozicioniranja mobilnog robota uporabom ultrazvučnog senzora i kompasa. Upravljački sustav mobilnog robota izgraĎen je uporabom Arduino platforme. Opisana je mobilna robotika sa naglaskom na podsustave potrebne za njegov rad. Detaljno su opisane sve komponente korištene pri izradi mobilnog robota. Opisano je područje umjetne inteligencije. Detaljno su opisane vrste učenja algoritama koji koriste umjetnu inteligenciju s naglaskom na Q-learning algoritam. Zatim je opisana izrada Q-learning algoritma, izrada sustava komunikacije za povezivanje mobilnog robota sa ureĎajima putem lokalne mreţe i izrada sustava za lokomociju mobilnog robota. Na kraju je izvršeno testiranje mobilnog robota i Q-learning algoritma. Testiranje je izvršeno na kvadratnom prostoru, a rezultati testiranja su opisani.This thesis suggests a method to build a mobile robot that uses artificial intelligence to find an optimal trajectory for the purpose of reaching the goal while avoiding obstacles. The thesis describes establishing the connection between a mobile robot and devices that use local area network. It also describes the process of positioning of a mobile robot using ultrasonic sensor and a compass. Control system was designed using Arduino platform. The thesis gives a general description of mobile robotics with emphasis on subsystems needed for it to work. This is followed with detailed description of each component used to build a mobile robot. The area of artificial intelligence was described. This is followed with a detailed description of types of learning with emphasis on Q-learning algorithm. Next, the description of designing the Q-learning algorithm, designing a system of communication used to connect a mobile robot to devices using local area network and designing of the system used for locomotion of the mobile robot. Finally, the mobile robot was tested, along with Q-learning algorithm. The test was done on square space. The results of the test were described