Crash statistics show that over 80% of motorist-bicycle crashes occur within 50 feet of traffic intersections. This paper focuses on the development of a collision avoidance system for bicycles for prediction and prevention of rear and side crashes at intersections. Cost, size and weight constraints highly limit the sensors and electronics that can be used on a bicycle and necessitate the development of new vehicle detection and tracking systems. Custom sonar and laser sensors and associated position estimation algorithms for tracking are developed. A custom sonar sensor with one sonar transmitter and two receivers is developed to estimate both the distance and angular orientation of vehicles on the sides of the bicycle. A custom single-target laser sensor on a rotating platform is developed to track longer distance vehicles. A model predictive control formulation is used to determine the real-time orientation of the rotating laser platform. Preliminary experimental data is presented to evaluate the performance of the side sonar system from a prototype instrumented bicycle. Simulation results are presented on the model predictive control for the rear vehicle tracking.