THE CLASS OF 2020 WILL BE DEFINED BY THE ATROCITY OF THIS...
SE PhD Prospectus Defense of Guang Yang
- 1:30 pm on Tuesday, August 20, 2019
- 3:30 pm on Tuesday, August 20, 2019
- 15 Saint Mary's Street, Rm 105
ABSTRACT: Motion planning is an essential part to many autonomous applications, such as self-driving car, unmanned aerial vehicle and humanoid robot. While many motion planning problems only limited to steering a dynamical system to a specific state while ensuring safety. For complex missions that involve time and logic constraints, a Temporal Logic (TL) guided motion planning is more suitable. In this research, the goal is to design motion planners for cyber-physical systems that satisfy the TL specifications using Barrier-certificate like approaches. We consider cyber-physical systems that utilize digital computers for both sensing and control implementation in discrete time but with continuous-time system dynamics. Due to the discrete time implementation of the controls, these systems often cannot guarantee the desired behavior in continuous time. In many cases, it can even cause potential safety hazard or inability to satisfy a given TL specification. To overcome the issue, we utilize the lower bounds of barrier constraints over time to ensure forward invariance in system states.
COMMITTEE:Advisor/Chair Calin Belta, SE/ME; Roberto Tron, SE/ME; Sean Andersson, SE/ME; Wenchao Li, SE/ECE