Implementation of Inverse Reinforcement Learning (IRL) algorithms in Python/Tensorflow. Deep MaxEnt, MaxEnt, LPIRL

Quadrotor control, path planning and trajectory optimization

Implementation of selected reinforcement learning algorithms in Tensorflow. A3C, DDPG, REINFORCE, DQN, etc.

Self-contained implementation of non-convex optimization algorithms in Python. Steepest Descent, Conjugate Gradient, Newton's Method, Quasi-newton (BFGS), l-BFGS

A full-fledged yet minimalistic CUDA-based convolutional neural network library from scratch in C++

Implemented Unscented Kalman Filter (UKF) for orientation tracking. Sensors fusion of accelerometer, and gyroscope

Implemented Gaussian Mixture Models (GMM) for image color segmentation.

Vision-based Robot 3D Poses and Velocities Estimation

Implemented a pipeline for automatic face replacement in videos (without deep learning). Image morphing, triangulation, TPS, Poisson blending, etc.

Particle-filter based SLAM.

Implemented Iterative Closest Point (ICP) method for 3d reconstruction

Path planning on high resolution aerial map based on imitation learning. Learn a cost-to-go map from demonstrations.

92% gender prediction accuracy based on users' tweets and profile pictures.

Implemented a pipeline for 2D image mosaic and stitching. Feature extraction & matching, Adaptive Non-Maximal Suppression (ANMS), geometric blur, RANSAC

Implemented face morphing algorithm based on Thin-Plate Splines

Implemented a Hidden Markov Model (HMM) for gestures recognition