Elyes Khechine

• Trained and fine-tuned RL models for time-optimal trajectory generation with the goal of reaching precise target positions and velocities while adhering to physical constraints, tailored for industrial packaging robots. • Reduced time discretization by 90% to achieve nearly time-optimal motion planning while maintaining task precision and compliance with motion constraints. • Designed and tuned reward systems to balance punishment factors, increasing task completion rates under tighter tolerances.

• Led the multi-disciplinary team through the technical development of 5 robotics projects, successful participation in 6 international competitions, and organization of 5 workshops, achieving 50% faster project completion with a tailored Crystal Agile approach. • Initiated the development of a Qt-based space robotics motion planning simulator, utilizing Docker for cross-platform support, and addressing challenges of ROS integration with DART, Gazebo, Unity, PyTorch, and SOLIDWORKS.

• Contributed to the development of 7 out of 42 AI projects, leveraging expertise in diverse deep learning and computer vision applications. • Participated in 70+ knowledge exchange sessions with 15+ AI enthusiasts. Discussed topics like machine learning, computer vision, and reinforcement learning, fostering a dynamic learning environment. • Led a technical workshop in computer vision and reinforcement learning for robotics, with a focus on policy gradient methods, Q-learning, and model-based approaches.

• Led a 6-member team to craft a system architecture for a 22kW 3-phase AC charging station prototype that meets client requirements and allows for full control over charging behavior and load-balancing in private and semi-public use cases. • Led the development of embedded C++ charging firmware with an ESP-WROVER-KIT board, ensuring precise measurement of charging current and voltage, over-the-air updates, and seamless TCP communication of events between the cloud server and the car.

• Advised project managers on product requirements for a 7.4kW 1-phase AC charging solution including key features, specifications, operating conditions, safety precautions, and functional requirements to meet client demands in the Tunisian and Netherlands markets. • Collaborated with 8 teammates to architect a high-level block diagram outlining electrical connections from the grid to the charging station and up to the EV, ensuring compliance with UL and IEC standards.

• Analyzed and presented novel robotic systems for manufacturing and assembly of 3D-printed solar arrays for electricity production in space. • Demonstrated innovative ways to utilize the generated electricity to power future spacecrafts through Ammonia fuel production with a PEM electrolyzer.