This dissertation describes an obstacle collision-avoidance technique that considers the dynamics of a vertical take-off and landing (VTOL) unmanned aerial vehicle (UAV). The dynamic characteristics and controller structures of VTOL UAVs, such as helicopters and tiltrotors, greatly vary depending on their flight envelopes. Previous research into rotorcraft collision avoidance mainly addressed the holonomic systems that consider hovering and low-speed flight. Nonetheless, a large rotary-wing UAV would still require collision-avoidance capabilities during take-off and landing, while the aircraft is in high-speed flight. This research proposes the Hybrid dynamic window approach (DWA) which combines the holonomic DWA, nonholonomic DWA and the flight dynamics constraints of a helicopter. The DWA method, which is similar to the MPC strategy but has small computational complexity, is utilized as the basic algorithm for motion planning with real-time applications in mind. Additionally, the generic two-dimensional dynamic window is extended to three dimensions to support forward flight, and a flight dynamics model is included to eliminate unreachable candidate paths. However, as the dimension of the search space expands, the amount of computation increases exponentially, and the size of the search space changes according to the constraints. Therefore, this research identifies the design parameters of Hybrid DWA suitable for rotorcraft, and proposes a design method to ensure the computational complexity, the effectiveness of solutions in the search space, and the rate of convergence on the target point. In order to verify the proposed scheme, various standard collision-avoidance scenarios are implemented, and the performance of Hybrid DWA is compared with the existing DWA method. The collision-avoidance performance is also verified through Monte-Carlo simulations that generated random obstacles. In conclusion, the proposed Hybrid DWA collision-avoidance technique is shown to be applicable to all flight-speed regimes of a typical rotorcraft, and it is demonstrated that the arrival probability of the target point is higher than that of the existing DWA algorithm - especially as the number of obstacles increases.