Publication
S. Dhakad, Design and Development of a Knee Rehabilitation Device for Hyperextension Prevention and Flexion Assistance, University of Maryland, College Park, [Technical Report], May 2025.
S. Dhakad, Design and Development of a Knee Rehabilitation Device for Hyperextension Prevention and Flexion Assistance, University of Maryland, College Park, [Technical Report], May 2025.
Developed a knee rehabilitation device to resist hyperextension and assist flexion via active and passive controllers.
Design 1: Passive controller, with Series Elastic Actuator (custom-designed torsional spring plate).
Design 2: Active Impedance PD Controller, BLDC Motor, Dynamic PD Gain Model.
Supervisor: Dr. Anindo Roy
Computer Vision Stack for horizon line detection, stop sign recognition, dynamic obstacle avoidance, and path navigation.
Implemented Hough Transform for vanishing point detection, trained a YOLOv8 model for stop sign identification, utilized optical flow techniques for real-time obstacle tracking, and employed contour-based methods for waypoint detection.
ROS2 Sim & Real-World Deployment: Closed-loop control system in ROS2 Humble, integrating all four perception modules to enable real-time decision-making.
Demonstrated robust autonomous navigation with TurtleBot3 in both simulated and real-world environments, achieving seamless coordination between perception and control.
Supervisor: Prof. Dr. Samer Charifa and Dr. Tommy Chang
Contributed to the development of the chassis, cargo compartment, and electro-mechanical actuation assemblies.
Designed jigs and fixtures for easy manufacturing and assembly (DFMA principles).
Implemented lean manufacturing with modular design of the main robot body.
Designed a drill tower for an autonomous inspection robot that can accommodate a 1 m hammer drill rod with an interchangeable drill setup. The design is sturdy towards mobile vibrations, topple conditions, and more.
Designed a semi-automatic screen-printing machine for small-scale industrial applications, where the printing process traditionally uses a manual screen-fabric frame. Developed the machine prototype using AutoCAD for CAD modeling and validated the mechanism operations through dynamic simulation in MechDesigner software.
Designed and structurally validated (CAD|CAE) four key mechanisms; Feed rollers for sheet pickup; Screen-frame actuator for controlled printing motion; Ink spreading mechanism for uniform ink distribution; Printed sheet ejection rollers for automated paper output.
Integrated a motor assembly with a timing gearbox to synchronize mechanical operations.
Designed custom PCB circuit to enable parameter setting and system control.
Supervisor: Prof. Dr. Sunil Punjabi
ROS-Based Autonomous Navigation: Implemented a navigation system on TurtleBot using ROS, enabling the robot to move autonomously through a predefined path.
ArUco Marker Integration: Utilized an onboard RGB camera to detect ArUco markers, which provided position-specific cues linked to parameters stored in a YAML file for guided movement.
Logical Camera Support: Integrated static logical cameras to enhance environmental perception and situational awareness, improving navigation reliability.
Modular ROS Package Development: Developed custom ROS nodes for marker detection, parameter retrieval, and navigation control using an action client, resulting in a robust, vision-guided navigation framework.
© 2024 Shivam Dhakad