Ankush Kadu Email & Phone Number

India

As a Research Intern at Ericsson, I am contributing to the development of explainability solutions for a 6G intent management system. This system utilizes Multi-Agent Reinforcement Learning (MARL) to manage intents such as throughput for various services. My work focuses on enhancing the transparency and trustworthiness of the system by providing clear, actionable insights into the decision-making processes of MARL. Through this, I aim to build trust and improve the system's reliability for real-world deployment.

Chennai, Tamil Nadu, India

Research Topic: Developing a Scalable Deep Reinforcement Learning Approach for COLREGs-Compliant Path Planning and Obstacle Avoidance in Autonomous Ships within dynamic maritime environments. Multi-Agent Deep Reinforcement Learning: Designing and implementing multi-agent reinforcement learning (MADRL) systems that allow multiple autonomous ships to navigate and coordinate in real-time, optimizing path planning and avoiding collisions based on dynamic environmental conditions. Explainable AI (XAI): Integrating explainability into the reinforcement learning models to provide transparent decision-making processes, ensuring that both human operators and regulatory bodies can understand and trust the autonomous ship's navigation choices, particularly in critical situations like collision avoidance. COLREGs Compliance Framework: Building a robust AI-driven framework that ensures navigation decisions strictly adhere to COLREGs (International Regulations for Preventing Collisions at Sea), maintaining compliance with established maritime safety standards. Safety and Collision Risk Mitigation: Prioritizing real-time decision-making to enhance maritime safety by reducing collision risks. This involves creating agents that learn optimal strategies to navigate through complex and changing maritime environments while dynamically responding to obstacles. Scalability for Large-Scale Operations: Focusing on the scalability of the deep reinforcement learning system to manage large fleets of autonomous ships, ensuring that the framework can be adapted to handle increasing numbers of agents without sacrificing efficiency or safety. Navigation Efficiency Enhancement: Utilizing advanced AI methodologies to improve navigation efficiency, optimizing routes not only for safe passage but also for energy conservation and time-sensitive operations, making the system both economically and operationally viable for large-scale deployment.

Chennai, Tamil Nadu, India

Chennai, Tamil Nadu, India

Navigation Module• Contributing to the development of Innoguide, an autonomous humanoid guide robot for personalized tours at the Center for Innovation (CFI) building (IIT Madras), employing ROS (Robot Operating System) for enhanced capabilities.• Technical Focus: Specializing in ROS-enabled navigation using machine learning algorithms for precise localization, mapping, and obstacle avoidance.• Key Features: Integration of advanced sensory equipment, voice communication, and human-robot interaction.• Ongoing Contributions: Leading advancements in the ROS-enabled navigation system to enhance precision in localization, mapping, and obstacle avoidance.

Master Of Science - Ms, Autonomous Marine Robotics, 9.12/10

Bachelor Of Engineering - Be, Mechanical Engineering, 9.36/10

Higher Secondary Education, 83%