In recent years, computer applications have significantly changed from traditional data processing to machine learning, due to the increased accessibility and availability of abundance of data gathered from the network and other resources. Machine learning automates human assistance by automating learning methods. In IoT integrated with Robotics, machine learning is applied to secure interconnected devices against cyber threats and insecurities. IoT is a network of interconnected gadgets and systems that exchange data over a network to perform various tasks using machines and robots. These machines range up to industrial machines and healthcare equipment which can be vulnerable to security threats. Machine learning is a prominent way to detect anomalies in real time. It enables predictive threat detection, continuous learning, and automated responses to improve the protection of IoT networks. In this paper, we explore the applications of various machine learning and its algorithms to strengthen the security of IoT devices, robotic machines and network. The paper presents the predictive ability of ML algorithms and their roles in improving threat detection and identification, reviewing their effectiveness against different types of challenges and vulnerabilities in IoT systems. Finally, the paper comprehensively reviews the current machine learning algorithms and their effectiveness in securing IoT network, devices, and robot while also offering some valuable insights and recommendations to create more adaptive and effective security solutions.