Rahul Mangharam (University of Pennsylvania): “Safety through Agility – Safe and Performant Control for Learning-Enabled Autonomous Systems”

Presentation Abstract:

We present three approaches to combine formal methods, control theory, and machine learning for safe and performant autonomous systems. 

1. Safe control for learning-enabled systems: We present our recent progress on how to learn safe adaptive behavior for highly interactive multi-agent systems. We will introduce how to quantify the uncertainty of closed-loop control systems using a frequentist method called conformal prediction and incorporate the uncertainty for safe perception-based control.
2. Learning Introspective Control: Oftentimes the systems that we control operate under different conditions due to changing environments, varying system parameters or changes in payload. As such, we strive to develop computationally efficient, data-driven system models that allow predictive controllers to adapt to changes in the environment in real-time. We focus on using Gaussian Processes as models to study the problem in the context of driving on surfaces with changing friction coefficients.
3. Differentiable Predictive Control: Finally, we discuss the application of differentiable predictive control for large-scale urban road networks.

Speaker Bio:

Rahul Mangharam builds safe autonomous systems at the intersection of formal methods, machine learning and controls. He applies his work to safety-critical autonomous vehicles, urban air mobility, life-critical medical devices, and AI Co-designers for complex systems. He is the Penn Director for the Department of Transportation’s $20MM Safety21 National Center. Rahul is the Director of the Autoware Center of Excellence for Autonomous Driving, a consortium of 70+ companies and universities focused on open-source AV software for open-standards EV platforms. Rahul received the 2016 US Presidential Early Career Award (PECASE) from President Obama for his work on Life-Critical Systems. 

This talk will be presented by Rahul’s doctoral students: Shuo Yang is a 3rd-year PhD candidate working on safe learning-enable controls, Ahmad Amine is a 1st-year PhD candidate working on learning introspective control, Nandan Tumu is a 4th-year PhD candidate working on physics-informed machine learning, all from ESE.