Charanpreet Narula Email & Phone Number

Burnaby, Bc, Ca

Burnaby, British Columbia, Canada

Dubai, United Arab Emirates

1) Software Architecture for Autonomous Robots: Engineered the software architecture using a combination of Adaptive Monte Carlo Localization and model predictive control to navigate and interact within urban environments.2) Advanced Perception and Interaction Capabilities: Enhanced the robot’s perception and interaction modules using a combination of deep learning models for face and emotion recognition, significantly improving human-robot interactions.3) Machine Learning Operations (MLOps) Implementation: Established an MLOps framework to facilitate continuous training, integration, and deployment of machine learning models, optimizing the robot’s performance in real-time.4) Intuitive Control Interfaces: Developed adaptive user interfaces using supervised learning algorithms to predict user preferences and customize controls, improving user experience and system accessibility.5) Sensor Integration and Data Fusion: Implemented unsupervised learning algorithms for sensor fusion, improving data accuracy from LiDARs, IMUs, and depth cameras for precise environmental mapping.6) Navigational Intelligence Enhancement: Advanced the robot's navigational capabilities using semi-supervised learning methods to process and interpret complex urban environments for safer and more effective advertising deployment.7) Developed Docking System to charge battery autonomously managed by a time scheduling system for continuous autonomous non-stop robot operations.8) Worked with a variety of sensors, 3D/2D lidar, Depth Camera, IMU, Ultrasonics integrated with Nvidia Jetson Orin and STM32 MCU.9) Worked on advanced algorithms for robust odometry, ICP Odometry, Sensor Fusion with Extended Kalman Filter/ Unscented Kalman Filter and VSLAM.10) Used Computer Vision and Deep learning models for object detection, pose detection and face expression analysis for AI Analytics.

Vancouver, British Columbia, Canada

1) Path Planning Innovation: Developed a custom path planning solution using A* and D* search algorithms, tailored to the robot’s specific operational needs for efficient navigation in crowded areas.2) Simulation and Model Testing: Leveraged reinforcement learning in simulation environments like Gazebo to dynamically adjust the robot’s decision-making processes based on simulated interactions and environmental changes.3) Cloud-Based Fleet Management: Integrated federated learning approaches to enhance the security and efficiency of fleet management systems, enabling decentralized data processing and real-time updates.4) Advanced Sensor Calibration and Navigation: Employed particle filter algorithms within the ROS2 framework to calibrate and synchronize multi-sensor inputs for accurate positioning and obstacle detection. 5) Worked on RTABMAP based Navigation using ROS2 and ROS1 bridge integrating with OpenRMF Fleet Management system.

Riyadh, Saudi Arabia

1) Developed a smart product identification API using state-of-the-art CNN architectures likeDarkNet53 and Yolo Models, implemented a distributed computing environment using Hadoop and Spark for handling large-scale data.2) Implemented an AI-driven facial recognition system for employee registration using deeplearning models and advanced face recognition techniques. Used High-PerformanceComputing for improved processing speed.3) Employed RNNs and transformer models for a Speech Recognition System, enhancing userinteraction with mobile apps. Worked with SQL and NoSQL databases for efficientdata management and extraction.4) Utilized Hive and Presto for efficient querying and managing big data stored in distributedsystems. Leveraged SparkSQL for data processing and analytics tasks on big data sets.5) Set up a robust MLOps pipeline using MLflow, Docker, and Jenkins, streamlining the machinelearning model lifecycle from development to deployment.

New Delhi, Delhi, India

1) Developed a cutting-edge lane detection system specifically designed for mobile robotics applications in dynamic and unpredictable environments. Using LSTM networks and complex sequence prediction algorithms, the system was adept at handling varied lighting and shadow conditions, significantly enhancing navigational reliability. The technology was pivotal in dynamically adjusting to environmental changes, ensuring consistent performance across different terrains and conditions.2) Development of a sophisticated perception system for mobile robots that integrated data from multiple sensors including LiDAR, cameras, and ultrasonic sensors. This system employed a combination of machine learning techniques to enhance the robot's environmental awareness and decision-making capabilities. By leveraging Extended Kalman Filters, we significantly improved the system's odometry and localization accuracy, which was crucial for navigating in open and complex environments.

Gurugram, Haryana, India

1) Autonomous Navigation for Cleaning Robots: Implemented SLAM algorithms with 3D LiDAR and RGB-D cameras to autonomously navigate and optimize cleaning paths for solar panel surfaces, significantly reducing manual oversight.2) Object Detection Using Deep Learning: Implemented CNNs to accurately identify and classify various objects on solar panels, improving the robot's effectiveness in debris removal.3) Precision Control with PID Algorithms: Utilized PID control techniques to precisely manage the movement of actuators in the cleaning apparatus, enhancing the robot's efficiency and operational stability.4) Development of Automated Assembly Systems: Applied deep reinforcement learning to automate the assembly processes using a YuMi-inspired robotic arm, which increased the assembly speed and precision by 40%.5) Quality Control with Machine Vision: Developed and trained object detection and segmentation models using TensorFlow to detect defects in auto parts, achieving a 95% accuracy rate in quality inspection and a significant reduction in operational costs.

New Delhi, Delhi, India

During my tenure with Team Defianz Racing, I was instrumental in developing and deploying an advanced data acquisition system tailored for Formula Student competition vehicles. This role required integrating and harmonizing data from a diverse array of sensors, including accelerometers, IMUs, strain gauges, and temperature sensors, through the Labjack DAQ system. The primary objective was to enhance real-time performance analytics, which was crucial for optimizing race strategies and vehicle settings under competitive conditions.

Master Of Engineering - Meng, Robotics And Control

MEng Final Project: Visual Inertial based State Estimation Using Landmark Detection

Bachelor Of Technology - Btech, Electrical, Electronics And Communications Engineering