The purpose of this research is to design the real-time fusion and signal processing algorithms to enhance the collision avoidance in autonomous vehicles. The study integrates the acquired data as well as a variety of sources like radar, LIDAR and cameras in order to form a particular impression of the surrounding of the car. Some of the sensor fusion techniques that are embraced are the Advanced sensor fusion such as the Extended Kalman filtering (EKF) and the Unscented Kalman filtering (UKF), an aided fusion of the different sensors and improves the accuracy on detecting objects. The signal processing lines, noise suppression and feature enhancement are used to improve sensor data to allow more dependable decision-making in dynamic and unpredictable driving situations. The system incorporates machine learning models whether as deep learning algorithms to anticipate possible collision aspects and activate relevant safety interventions, including automated emergency braking and collision avoidance. The proposed system was proven to be effective through simulations in MATLAB that revealed great improvement in detection, increased response time, and stability in the system to be used in different real-world situations. The results underline the prospect of these methods making a considerable contribution to the vehicle safety in the case of autonomous driving implementation to the further increase in reliability and decrease in the risk of accidents in the situation of reality.