Stanford, California, United States
Employed and optimized a model predictive control-based driver model on a high fidelity racing simulator. Progress included developing a smaller scale simulator with simplified vehicle dynamic models for quick simulation runs, tuning the MPC quadratic cost function gains via sophisticated optimization techniques, and deep learning approaches were implemented to further enhance controller performance.
An additional previous project involved implementing decision making and planning techniques for an autonomous vehicle performing multiple lane changes in an urban environment, in which the problem was formulated as a partially observable Markov decision process.
These projects have been done with the Stanford Intelligent Systems Laboratory.
