Bipedal Walking can be characterized by a sequence of foot steps with intermediate impacts with the ground. Different contact situations with the ground may arise, for example a roll around the heel of the supporting foot yielding an underactuated ballistic motion phase. The aspects of underactuation, non-neglectable impacts with the ground, and unilateral contact with the ground make bipedal walking a challenging and interesting problem for trajectory planning and control.

The aim in the control of bipedal walking is to achieve dynamic, energy-efficient and stable walking motions comparable to human walking performance. Although a human-like walking style is desirable, recorded human walking trajectories are not simply to be duplicated. Instead, the underlying fundamental principles shall be identified and then used to create the bipedal robot's walking trajectories. Currently investigated principles include the contact situation sequence of the gait, optimal control, and elasticities at the joints.

A topic of current research is the extension of an existing trajectory planning framework based on so-called "virtual holonomic constraints" to the case of double support. In double support, a redundancy in the configuration variables arises and additional contact forces have to be considered. With the inclusion of double support, more complex foot contact sequences can be chosen for walking gaits and then compared with respect to energy consumption and similar objectives.

One promising approach to increase energy efficiency of walking is the use of elasticities. This approach is inspired by the elasticities in the human muscle-tendon system. Optimal control is used to optimize gaits with and without additional springs at internal joints. A comparison shows, that with springs a significant reduction of energy consumption can be achieved, along with a more human-like gait.