Events

PhD Defense: "Quantification and Optimization of Robustness for Two-Legged Robots Walking on Rough Terrain"

Cenk Oguz Saglam

June 9th (Tuesday), 2:00pm
Elings Hall, Room 1605


For practical and autonomous operation in real-world environments, two-legged robots need optimization of performance using meaningful metrics. Speed and energy efficiency are straightforward to quantify, but robustness must also be measured for reliability under variable or otherwise uncertain environmental conditions including rough terrain. The intuitive and meaningful robustness quantification adopted in this thesis begins by stochastic modeling of the disturbances such as terrain variations, and conservatively defining what a failure is, for example falling down, slippage, scuffing, stance foot rotation, or a combination of such events. After discretizing the disturbance and state sets by meshing, step-to-step dynamics are studied to treat the system as a Markov chain. Then, the failure rates can be easily quantified by calculating the expected number of steps before failure. Once the robustness is measured, other performance metrics can also be easily incorporated into the cost function for optimization.

In addition to optimization of blind-to-the-environment controllers, an intuitive and capacious approach to maximize performance of legged robots is to adopt a hierarchical control structure. Given environment estimation and state information, the high-level control is a behavioral policy to choose the right low-level controller at each step. In this thesis, optimal policies are determined by applying dynamic programming tools on Markov decision processes that are already obtained to calculate the expected number of steps for reliability measurement. Robustness of high-level control to environment estimation and discretization errors are ensured by modeling stochastic noise in the terrain information and belief state while solving for behavioral policies.

About Cenk Oguz Saglam:

photo of Cenk Oguz Saglam Cenk Oguz Saglam received the B.S. degree in mechatronics engineering from Sabanci University, Turkey, during which he was a visiting student in University of California, Los Angeles, and a research intern at NanoRobotics Laboratory of Carnegie Mellon University. He received his M.S. degree in electrical and computer engineering from University of California, Santa Barbara, where he is currently a Ph.D. candidate. His current research interests lie primarily on modeling, analysis, and control of highly-dynamic robots in real-world environments, which often imply underactuation and stochasticity. In particular, he is exploring methods for incorporating human-like walking dynamics into bipedal robots on rough terrain for robustness and autonomy. Oguz has an interdisciplinary background and has worked on bilateral control systems and wall-climbing robots before.

Hosted by: Professor Katie Byl, Robotics Lab