Vehicle collisions continue to be a major problem for transportation systems, resulting in a considerable loss of life and property. Collision avoidance strategies have become more popular since the introduction of connected and autonomous cars. In vehicular ad hoc networks (VANETs), collision avoidance is essential for maintaining congestion and road safety. This paper presents a comprehensive framework for vehicle collision avoidance integrating K-means clustering and a threshold-based decision-making mechanism. The technique ensures localized and effective communication within clusters by dynamically grouping vehicles into clusters depending on attributes like position and distance. For safe distance among vehicles, threshold has been introduced reducing the risk of collision by prompting an alert to the vehicles to take corrective measures when risk exceeds the safety threshold. Simulation results validate the effectiveness of the proposed mechanism in mitigating collision risks under varying traffic densities and mobility patterns. The integration of K-means clustering reduces network overhead by minimizing redundant inter-vehicle communication, while the threshold-based system ensures decision-making for safety–critical scenarios.