Autonomous vehicles are an integral component of intelligent logistics in smart cities. Accurate navigation and reliable obstacle detection are key features of these vehicles. The use of appropriate and advanced sensors is essential to achieve these capabilities. In the previous work, different sensors perform their sensing independently. This study proposes a solution for optimal navigation and obstacle detection using a low-cost, dual-mode, sensor-fusion-based navigation system (comprising infrared and ultrasonic sensors), enabling the vehicle to follow pre-programmed paths. Additionally, a smartphone app can help drivers maneuver the vehicle in both manual and automatic control modes. The advantages of this smart navigation lie in reduced cost due to simple and economical components, that results into layered control architecture, reproducible, feasible and scalable solution. Further, the obstacle detection efficiency is improved to be 95%. Preliminary experiments indicate the technique has high accuracy in navigation and obstacle detection and is capable of lessening the level of human reliance and improving the reliability of delivery.