The existing public transport system in Addis Ababa is critically insufficient to provide servicernfor the existing travel demand. Public transportation is one and the most important means ofrneasing traffic congestion for it makes roads work better by reducing the number of vehicles onrnthe road. This shows that a great concern should go towards the expansion of high- capacityrnpublic transportation system like introduction of light rail, heavy rail, road bus transit, and highrnoccupancy vehicles lanes, which coupled with better management of the existing road networkrnand traffic management. We need therefore to have a clearly defined transport managementrntechnology to challenge the mobility issues of the city. Grade Crossing is a location where arnpublic highway, road, street, or private roadway, including associated sidewalks, and pathways,rncrosses railroad tracks at grade (same level as the street).The aim of this thesis is to encouragerngrade crossing safety and reduce highway traffic delay. The safety of grade crossing can bernpromoted by removing those vehicles detected on the railroad tracks before the arrival of trainsrnusing CCTV camera by image processing technology. Reduce highway traffic delay by properrnmanaging the phase sequence of the intersection. A system model is proposed and developingrnwith the system. The optimization will be implemented into two steps. The first step, the delayrnfunction is approximated and represented by artificial neural network. Secondly optimizationrnwill be applied based on Levenberg-Marquardt optimization.rnWe use different approach, instead of protecting grade crossing manually and showing thernhostility to the highway drivers, we could develop an advanced traffic management system smartrnenough to control the traffic near grade crossings. In such a system, we could incorporate graderncrossing information into traffic control and prevent the queue from backing onto the railroadrntracks. Preemption of the traffic signal at/near a grade crossing is such an alternative to targetrnsafety improvement. Artificial neural network is design for both train arrival time at graderncrossing and forecasting of traffic signal phase length at intersection. MATLAB programing isrnapplied for the designs simulation. From the simulation result the network is optimized byrndecreasing MSE of train arrival time and traffic signal phase length prediction by 2.2696x10-19rnand 8.338x10-9 respectively. Image processing is applied for obstacle detections on graderncrossing by the concept of image segmented in matrix form.