Autonomous surface craft (ASC) are increasingly attractive as a means for performing harbor operations including monitoring and inspection. However, due to the presence of many fixed and moving structures such as pilings, moorings, and vessels, harbor environments are extremely dynamic and cluttered. In order to move autonomously in such conditions ASC's must be capable of detecting stationary and moving objects and plan their paths accordingly. We propose a simple and scalable online navigation scheme, wherein the relative motion of surrounding obstacles is estimated by the ASC, and the motion plan is modified accordingly at each time step. Since the approach is model-free and its decisions are made at a high frequency, the system is able to deal with highly dynamic scenarios. We deployed ASC's in the Selat Pauh region of Singapore Harbor to test the technique using a short-range 2-D laser sensor; detection in the rough waters we encountered was quite poor. Nonetheless, the ASC's were able to avoid both stationary as well as mobile obstacles, the motions of which were unknown a priori. The successful demonstration of obstacle avoidance in the field validates our fast online approach.