Shankar Mohan received a B.Eng from the National University of Singapore (2011), and his M.S. and PhD. from the University of Michigan (2013, 2017), all in Electrical Engineering. Since March 2017, he has been employed by Ford Motor company as a Research Engineer, and works on applying optimization based techniques to problems in robotic and automotive systems. His research interest spans energy storage systems, power systems, numerical methods for optimization and optimal control problems, and safety and reliability of autonomous systems.

Planning and executing robust & safe maneuvers is of critical importance for autonomous/robotic systems. Addressing such problems is made difficult by the fact that global optimal solutions to nonlinear optimization problems are difficult to obtain. We have developed implementable numerical techniques that generate provably-consistent approximations of the global optima for such problems.

• Moment relaxations/truncation was shown to be useful in estimating probabilistic safe sets of nonlinear hybrid systems such as biped robots with parametric uncertainty
• A method to generate optimal policies to constrained nonlinear optimal control problems on hybrid systems was developed
• A novel controller-in-the middle structure has been developed to change the behavior of a class of systems driven by optimal policies
• Developed safety controllers to support RL based decision makers

The proliferation of distributed power sources and nonlinear loads has warranted better control of grid-interfacing power electronic systems. The following are some of my relevant work in this domain:

• Developed a novel power-flow and control strategy aimed at increasing the power density and reducing the cost of inverters connected to unbalanced micro-grids with asymmetric faults and nonlinear loads
• Designed a novel communication-free voltage and frequency stabilization of microgrids using injected sub-harmonic signatures
• Assessed the economic benefit of grid-scale Battery Storage Systems when participating in electric markets, to gauge market potential

Li-ion batteries suffer from significant performance degradation (in terms of power capability) when operating from sub-zero temperatures. Work on this problem culminated in the following outcomes:

• Proposed a novel and easy-to-tune nonlinear estimator structure that leverages the notion of relative observability
• Devised a method to certify the existence of (or lack thereof of) an energy-efficient battery warm-up strategy, empowering mission planning systems
• Proved that conventionally applied techniques for battery warm-up are near optimal
• Developed a novel technique to warm-up batteries; this method reduces energy consumption by ~20% over conventional methods