Autonomous driving systems need resilient real-time perception and path-planning capability to guarantee secure navigation in dynamic conditions. This study introduces an autonomous driving assistance system that combines YOLOv3 with the Rapidly-exploring Random Tree (RRT) algorithm on the NVIDIA Jetson Orin Nano platform. YOLOv3, an object identification model based on deep learning, analyzes camera pictures to recognize and categorize road obstacles, such as automobiles, pedestrians, and traffic signs. Developed in Python, YOLOv3 is tuned for efficient operation on the Jetson Orin Nano, guaranteeing real-time performance. The system assesses the distances of identified objects and utilizes the RRT algorithm to create a secure course that circumvents any dangers. The RRT algorithm investigates several random trajectories and identifies the ideal route depending on parameters such as distance, safety, and vehicle agility. Experimental findings indicate that the suggested system proficiently identifies impediments and formulates collision-free trajectories in real-time, hence improving driving safety. This study advances autonomous driving and intelligent transportation systems, facilitating safer and more efficient road transportation in the future.