本文简要介绍了美国国防部先进研究项目局举办的机器人大挑战赛的历史，并将关注的焦点放在了2012机器人挑战赛上。作为该挑战赛DRC-HUBO队的成员，我们为完成崎岖地面的任务提出了不同的方法，如增大脚踏板、转换成四足行走等。同时，本文还介绍了一种帮助提高仿人机器人运动稳定性的新步态——滑雪式步态。本文对该步态的稳定性进行了分析，并利用稳定度裕量从两种备选的步进顺序（爬行-1和爬行-2）中进行选择。接下来，本文对滑雪式步态中冗余的闭链系统进行力/力矩分析，通过最小化总耗能量来确定关节力矩。根据稳定性分析和力/力矩分析，我们设计了一种手杖长度，使其可以支撑HUBO2仿人机器人平台稳定地实现爬行-2步态。最后，通过将本文的试验结果和两足步行进行对比，验证了滑雪式步态的可行性。同时，介绍了我队在该机器人挑战赛上的表现。

In this paper, we briefly introduce the history of the Defense Advanced Research Projects Agency (DARPA) Grand Challenge programs with particular focus on the 2012 Robotics Challenge. As members of team DRC-HUBO, we propose different approaches for the Rough-Terrain task, such as enlarged foot pedals and a transformation into quadruped walking. We also introduce a new gait for humanoid robot locomotion to improve stability performance, called the Ski-Type gait. We analyze the stability performance of this gait and use the stability margin to choose between two candidate step sequences, Crawl-1 and Crawl-2. Next, we perform a force/torque analysis for the redundant closed-chain system in the Ski-Type gait, and determine the joint torques by minimizing the total energy consumption. Based on the stability and force/torque analysis, we design a cane length to support a feasible and stable Crawl-2 gait on the HUBO2 humanoid robot platform. Finally, we compare our experimental results with biped walking to validate the Ski-Type gait. We also present our team performance in the trials of the Robotics Challenge.