A Stability-Estimator to Unify Humanoid Locomotion: Walking, Stair-Climbing and Ladder-Climbing

The field of Humanoid robotics research has often struggled to find a unique niche that is not better served by other forms of robot. Unlike more traditional industrials robots with a specific purpose, a humanoid robot is not necessarily optimized for any particular task, due to the complexity and balance issues of being bipedal. However, the versatility of a humanoid robot may be ideal for applications such as search and rescue. Disaster sites with chemical, biological, or radiation contamination mean that human rescue workers may face untenable risk. Using a humanoid robot in these dangerous circumstances could make emergency response faster and save human lives. Despite the many successes of existing mobile robots in search and rescue, stair and ladder climbing remains a challenging task due to their form. To execute ladder climbing motions effectively, a humanoid robot requires a reliable estimate of stability. Traditional methods such as Zero Moment Point are not applicable to vertical climbing, and do not account for force limits imposed on end-effectors. This dissertation implements a simple contact wrench space method using a linear combination of contact wrenches. Experiments in simulation showed ZMP equivalence on flat ground. Furthermore, the estimator was able to predict stability with four point contact on a vertical ladder. Finally, an extension of the presented method is proposed based on these findings to address the limitations of the linear combination.Ph.D., Mechanical Engineering and Mechanics -- Drexel University, 201