A key component of robotics, robot navigation helps autonomous robots to navigate effectively and securely throughout their environment. Conventionally, programmers and engineers have guided robot motions using set guidelines and heuristics. These guidelines, however, may apply only to particular settings and have difficulty adjusting to new or changing circumstances. For basic path planning algorithms like obstacle avoidance, for example, are often utilized. Effective in controlled environments, they lack the adaptability required to negotiate varied and erratic situations. Machine learning (ML) has now become a quite interesting substitute. ML lets robots use real-time sensory data to learn from it and modify their navigation plans. Thus, with an eye toward more competent and flexible robotic systems, this study concentrates on using ML for robot navigation categorization. This method helps robots to learn from their experiences and sensory inputs, therefore enhancing their capacity to negotiate difficult surroundings. Beyond what conventional rule-based approaches can manage, this adaptable technique is especially helpful in situations when the surroundings change often or provide varied and difficult hurdles. By using ultrasonic sensor data as input, the robot gains important distance information that helps it precisely recognize and avoid obstacles. Moreover, including ML into robot navigation improves their capacity to manage changing surroundings and challenging real-world situations. Using ultrasonic sensor data turns out to be a wise decision as it offers important data for precise path planning and obstacle identification. In the end, this suggested ML-based approach emphasizes the possibilities of ML methods (i.e., logistic regression, and multilayer perceptron) in improving robot navigation capabilities, so opening doors for more advanced and autonomous robotic systems able of running effectively in various and unpredictable environments.