Ing. Muhammad Usman Khalid

Responsible for improvement of current bin picking software by using deep learning and computer vision techniques. Implementation of algorithms for object segmentation of bin and work pieces, object detection of work piece and 6D pose estimation of work piece. Implementation of control software for Schunk Gripper.

• Proposed and Implemented Deep Learning approach, which used estimated human pose directly as an input to Convolutional Neural Network for action recognition. • Implemented Pose-based CNN in Keras (Tensor flow) and achieved state-of-art results on benchmark datasets. • Improved the accuracy of algorithm by applying different fusion schemes on multi-modal information (Pose, RGB stream and optical flows). • Implemented also Conventional Machine Learning Approach and made a comparison of results.

• Prepared large data by extraction of action based video segments (RGB, Optical flows, Pose) for each person in video for Bosch internal dataset. • Using OpenCV library, extracted hand crafted features from videos and using Bag-of-words approach, trained Conventional Machine Learning Approach like SVM to do action classification. • Successfully implemented a live demo for the visualization of results for action recognition.

• To Implement the DAE model in Python for the BASF Reactor for ongoing European Research Project MOBOCON. • Use parallel programming in Python to realize the optimizer in BASF Reactor. • To Implement NLP (Non Linear Programming) for Optimal Control of BASF Reactor using Python.

• Develop cost effective optimized algorithms to design a complete transformer. • Using MATLAB, redefined cost effective optimized modules based on daily rates of Copper and Silicon Steel.

• Develop modules using Python and integrated to PSSE for the complete feasibility analysis of power systems. • Developed PSSE integrated Python routines for doing the recursive tasks automatically in Load flow and contingency analysis. • Developed PSSE integrated Python modules that made the base cases for power system analysis for all 12 months of the year automatically.

Machine Learning in Robotics • Statistics for Researchers • 3D Computer Vision • Computational Intelligence • Modelling and Control of Robotics Manipulators Mobile Robots • Control Systems • Fundamentals of Robotics • Logic Controls • Computer Systems • Advanced Engineering Mathematics

Control Theory, Power Systems, Renewable Energy Sources, High Voltage Engineering, Power Electronics