Design and Research of Reinforcement Learning Algorithms and Software for a Robotic Ground Platform

This thesis presents the justification of the need to develop a universal robotic platform for environmental monitoring and safety at industrial facilities. The relevance of the work is due to the growing number of industrial enterprises, hazardous production facilities and the risk of accidents, poisoning and environmental disasters. Existing monitoring systems are not always effective in complex and dynamic environments, require constant maintenance and human participation, which creates health and safety risks. The goal of the work is to create a robotic platform capable of autonomously performing monitoring tasks for various environmental parameters, learning and adapting to changing conditions. The work will investigate reinforcement learning algorithms used to control the platform. The research will not be conducted in virtual environments, but on a physical ground-based robotic platform. Expected results: Development of an effective robotic platform for environmental monitoring and safety. Reduction in the number of accidents and poisonings. Increased safety of people and the environment.