As unmanned ground vehicle technology matures and autonomous platforms become more common, such platforms will invariably be in close proximity to one another both in formation and independently. With an increasingly crowded field, the risk of collisions between these platforms grows, and with it the need for path deconfliction. This paper presents two complementary technological developments to this end: a pipeline for affirmatively identifying and classifying dynamic objects, e.g., vehicles or pedestrians; and a pipeline for preventing collisions with such objects. The efficacy of these techniques is demonstrated in simulation, and validation on robotic platforms will be undertaken in the near future.