In the age of advanced technology, the deployment of mobile robots has emerged as a vital solution, transforming various industries by automating tasks, improving efficiency, and shaping the future of autonomous systems. One significant advancement in the past decade is Reinforcement Learning, made possible by the evolution of deep learning, allowing for task generalization and impressive results. However, modern Reinforcement Learning algorithms still rely on a trial-and-error approach, leading to high costs in terms of time and money. This thesis explores an alternative strategy by employing Offline Reinforcement Learning algorithms in these tasks. It demonstrates instances where this new approach yields positive outcomes and enables purely offline training, saving both time and costs.