This chapter presents a kinodynamic motion planner for computing agile motions of quad-rotor-like aerial robots in constrained environments. The approach uses a local trajectory planner based on fourth-order splines and a simplified model of the UAV to generate flyable trajectories of minimal time. The local planner is integrated into a global motion planning framework using different approaches. The performance of the proposed methods is demonstrated through simulation results and preliminary experiments with a real quadrotor. The chapter discusses a decoupled approach for motion planning, which involves first planning a geometrically valid path for the minimum bounding sphere of the quadrotor and then transforming this path into a dynamic trajectory. This approach is computationally efficient and can be applied to solve many motion planning problems for UAVs. However, it fails to handle certain problems, such as the slot problem, where the robot must navigate through a narrow passage smaller than its bounding sphere. Another type of problem that cannot be addressed by the decoupled approach is the transportation of a large object. In such cases, the minimum bounding sphere is not suitable, and the actual shape of the system must be considered for collision detection. The chapter also presents a direct kinodynamic planning approach, which uses sampling-based methods to generate trajectories directly, without first planning a geometric path. The chapter introduces a quasi-metric in the state space that accurately estimates the cost-to-go between two states and an incremental sampling technique that increases the probability of generating connectible states. These techniques significantly improve the performance of the PRM-based and RRT-based planners, particularly for solving very constrained problems. Simulation results show that the integration of these techniques significantly improves the performance of the planners. The chapter also presents preliminary experimental results with a real quadrotor, demonstrating the ability of the planner to generate trajectories that can be executed by a real quadrotor. The results show that the quadrotor can follow the planned trajectory, although some tracking errors were observed at high speeds. Future work includes further experimental validation and improvements in localization and control techniques to reduce tracking errors and demonstrate the ability to execute more agile maneuvers in constrained environments.This chapter presents a kinodynamic motion planner for computing agile motions of quad-rotor-like aerial robots in constrained environments. The approach uses a local trajectory planner based on fourth-order splines and a simplified model of the UAV to generate flyable trajectories of minimal time. The local planner is integrated into a global motion planning framework using different approaches. The performance of the proposed methods is demonstrated through simulation results and preliminary experiments with a real quadrotor. The chapter discusses a decoupled approach for motion planning, which involves first planning a geometrically valid path for the minimum bounding sphere of the quadrotor and then transforming this path into a dynamic trajectory. This approach is computationally efficient and can be applied to solve many motion planning problems for UAVs. However, it fails to handle certain problems, such as the slot problem, where the robot must navigate through a narrow passage smaller than its bounding sphere. Another type of problem that cannot be addressed by the decoupled approach is the transportation of a large object. In such cases, the minimum bounding sphere is not suitable, and the actual shape of the system must be considered for collision detection. The chapter also presents a direct kinodynamic planning approach, which uses sampling-based methods to generate trajectories directly, without first planning a geometric path. The chapter introduces a quasi-metric in the state space that accurately estimates the cost-to-go between two states and an incremental sampling technique that increases the probability of generating connectible states. These techniques significantly improve the performance of the PRM-based and RRT-based planners, particularly for solving very constrained problems. Simulation results show that the integration of these techniques significantly improves the performance of the planners. The chapter also presents preliminary experimental results with a real quadrotor, demonstrating the ability of the planner to generate trajectories that can be executed by a real quadrotor. The results show that the quadrotor can follow the planned trajectory, although some tracking errors were observed at high speeds. Future work includes further experimental validation and improvements in localization and control techniques to reduce tracking errors and demonstrate the ability to execute more agile maneuvers in constrained environments.