In dynamic uncertain environment, the security and reliability of UAVs executing missions can be improved greatly by taking the mobility of the threatening obstacle and the detection error of sensors onboard the UAVs into consideration. On the basis of 3-D velocity obstacle model, the dynamic uncertainties is expressed by the orientation error of threatening obstacle's velocity, and the 3-D Dynamic Velocity Obstacle Model (3-DDVOM) is established. A 3-D dynamic uncertain autonomous collision avoidance algorithm based on 3-DDVOM is presented. Finally, the proposed 3-D autonomous collision avoidance method for UAV is applied to online path planning of Pythagorean Hodograph (PH) curve, and the simulation results show the effectiveness and feasibility of the method.