An Extension to Model Predictive Path Integral Control and Modeling Considerations for Off-road Autonomous Driving in Complex Environment - Robotics Institute Carnegie Mellon University
Loading Events

PhD Speaking Qualifier

September

8
Thu
Charles Noren PhD Student Robotics Institute,
Carnegie Mellon University
Thursday, September 8
1:30 pm to 2:30 pm
NSH 3305
An Extension to Model Predictive Path Integral Control and Modeling Considerations for Off-road Autonomous Driving in Complex Environment

Abstract: 

The ability to traverse complex environments and terrains is critical to autonomously driving off-road in a fast and safe manner. Challenges such as terrain navigation and vehicle rollover prevention become imperative due to the off-road vehicle configuration and the operating environment itself. This talk will introduce some of these challenges and the different tools developed to handle the difficulties of off-road driving. Specifically, our key contributions are an extension of the state-transition model used in Model Predictive Path Integral (MPPI) Control to handle changes in terrain elevation via a modeling of the vehicle’s suspension; and a first principles approach to developing vehicle roll limits for use in rollover analysis or online motion planning. I will present results from both simulation and from hardware runs that reflect how our extension to the MPPI controller transition model influences the vehicle control decisions to account for gravitational acceleration as the vehicle traverses complex terrains. Additionally, I apply our rollover model to two different vehicle rollovers experienced during testing –one reported and one recorded- that show that rollover occurs in agreement with the rollover limits established by our model.

Committee Members: 
Prof. Matthew Travers
Prof. Sebastian Scherer
Prof. Michael Kaess
Brian Jackson