A fast nonlinear biped gait generator with state constraints

How to quickly and reliably solve constrained nonlinear optimization problems remains an active research topic in bipedal robot gait planning. To address this issue, this paper presents a new nonlinear gait generation framework that incorporates the Augmented Lagrangian–Differential Dynamic Programming (AL-DDP) algorithm. The integration of AL-DDP not only enables efficient handling of various types of constraints but also meets the real-time requirements of online gait generation. Moreover, unlike some previous approaches that only consider control constraints, this paper can explicitly incorporate state constraints into the planning process. This improvement allows the robot to maintain feasible motions in height-restricted environments and facilitates the execution of more tasks. Finally, to validate the effectiveness of the proposed framework, simulations and experiments are conducted on the Nao-v5 humanoid robot. The results show that this framework enables the robot to achieve stable walking under various conditions.