Sabeeh ul hassan Saad

-Navigation of a car in a multi-agent, multi-lane environment using deep reinforcement learning -For this task, an approach is designed that can tackle continuous states as an input -For simulation and training, Simulation for Urban Mobility (SUMO) is used -Approach is implemented in python using Tensorflow and TRACI API to interact with SUMO -Results show that the car is able to navigate successfully due to the proposed approach

-Design of an approach to make Dominion software (Simulation software for autonomous cars internally used in DLR) compatible with CoinCarSim( an opensource ROS based framework) -For this, a socket-client based interface was designed by me that makes the compatibility between the two softwares possible -This framework was later on used in real time to drive Volkswagen car at DLR

-Wrote complete spiral based path coverage algorithm and path covered marking for industrial vacuum cleaner robot named as “Roxxter” made by Bosch. User can also visualize the path covered by the vacuum cleaner and control it via mobile app -Computer Simulations of the algorithm with the help of ROS, gazebo -Item recognition while putting in the fridge so that the user knows what is inside the fridge, used deep learning approach-YOLO -Sound navigation of Bosch-robot with the help of ultrasonic sensors

-Electronics-hardware for a patented design -Motion tracking with the help of different deep learning architectures such as YOLO, SSD -Teaching of basic concepts of machine, deep learning to the team so that they can work on their own -Computer vision tasks related to patented design as well as other projects -GUI's in Visual studio and Qt for experimentation -Technical documentation

-Support of scientific staff on a European Union funded robotics project -Hardware and software development of the proposed robot which includes working on different microcontroller, electronics and fast charging systems for the proposed robot -PCB designing using ALTIUM designer to be used for lifting mechanisms -Creation of experimental setups for the robot -Conducting field tests for the robot -Making 3D drawings-animations in Inventor, Solidworks and 3Ds Max

-Completed ROS based speech-text module, later on used on Jenny(humanoid) in Robocup competition -Sound direction recognition so that the robot can find out the direction of sound coming. Once detected, robot moves towards the direction of the sound

Robotics, Machine Learning, Deep Learning, Robot Navigation, Robot Manipulation, Software and hardware design