Wheeled-legged robots are an attractive solution for versatile locomotion in challenging terrain. They combine the speed and efficiency of wheels with the ability of legs to traverse challenging terrain. In general, the challenges with wheeled-legged locomotion involve trajectory generation and motion control for trajectory tracking. This thesis focuses in particular on the trajectory optimization task for wheeled-legged robots navigating in challenging terrain. For this, a motion planning framework is proposed that optimizes over the robot’s base position and orientation, and the wheels’ positions and contact forces in a single planning problem, taking into account the terrain information and the robot dynamics. The robot is modeled as a single rigid-body, which allows to plan complex motions for long time horizons and still keep a low computational complexity to solve the optimization quickly. The knowledge of the terrain map allows the optimizer to generate feasible motions for obstacle negotiation in a dynamic manner, at higher speeds. Such motions cannot be discovered without taking into account the terrain information. Two different formulations allow for either purely driving motions, where obstacle negotiation is enabled by the legs, or hybrid driving-walking motions, which are able to overcome discontinuities in the terrain profile. The optimization is formulated as a Nonlinear Programming Problem (NLP) and the reference motions are tracked by a hierarchical whole-body controller that computes the torque actuation commands for the robot. The trajectories are verified on the quadrupedal robot ANYmal equipped with non-steerable torque-controlled wheels in simulations and experimental tests. The proposed trajectory optimization framework enables wheeled-legged robots to navigate over challenging terrain, e.g., steps, slopes, stairs, while negotiating these obstacles with dynamic motions.