Samriddhi Dubey

BTech Projects

Analyzing the Impact of Increasing Leakage Fault in Single and Multiple Cylinders and Pressure Fluctuation Reduction via Piston Rearrangement in an Axial Piston Pump

2024 Prof. Pavan Kumar Kankar IIT Indore

Key Innovation

Discovered that uniform piston arrangement produces the least pressure fluctuation in faulty systems, providing a novel approach to fault management in axial piston pumps.

This project was carried out during my final year of BTech at IIT Indore under the guidance of Prof. Pavan Kumar Kankar in the Department of Mechanical Engineering. It focuses on the detailed analysis of failure modes in axial piston pumps (APPs), which are commonly used in industrial systems, automotive applications, and manufacturing setups. Despite their efficiency, these pumps face significant performance challenges due to internal leakage and pressure instability.

View Full Project Report

Leakage flow rate

Outlet Pressure

Methodology

The core of the work involved developing a simulation model using the SIMSCAPE environment in MATLAB Simulink. A closed-loop variable displacement axial piston pump was modeled, accounting for the clearance between the piston and the cylinder block.

01

Simulation Development

Created a detailed model in MATLAB Simulink with variable clearance parameters

02

Fault Analysis

Simulated leakage faults by varying annular clearance from 1µm to 100µm

03

Validation

Compared results with experimental data from a Vickers PVB5 pump

04

Piston Arrangement Study

Analyzed inline, random, and uniform configurations for optimal performance

Key Findings

Fault Progression

Identified three distinct stages of degradation in pressure signals as leakage severity increases

Performance Impact

Increasing leakage reduced flow rate and disturbed outlet pressure with measurable patterns

Optimal Configuration

Uniform piston arrangement produced the least pressure fluctuation in faulty systems

Piston Inline Arrangements

Piston Random Arrangements

Piston Uniform Arrangements

Conclusion & Impact

This research contributes valuable insights into optimizing the performance, reliability, and maintenance of axial piston pumps in practical settings. The findings on piston arrangement strategies provide a novel approach to managing pressure fluctuations in systems with leakage faults, potentially extending equipment life and reducing maintenance costs in industrial applications.

SPAC-R: Suction-Driven Paper Assembly and Crafting Robot

MTech,1st semester IIT Gandhinagar Delta Robot

Key Innovation

A delta robot with suction-based end effector designed for precise paper manipulation and assembly tasks, enabling automated crafting and paper-based construction.

I completed this project during my 1st semester's Mechatronics course at IIT Gandhinagar under the supervision of Prof. Madhu Vadali. SPAC-R is a specialized delta robot designed for precise paper manipulation and assembly. The robot utilizes a suction-based end effector to pick, place, and manipulate paper components with high precision, enabling automated crafting and paper-based construction tasks.

View on GitHub

Project Demo

Watch SPAC-R Demo on YouTube

Methodology

The development of SPAC-R involved designing a delta robot architecture optimized for paper manipulation, with a focus on precision movement and gentle handling through suction-based gripping.

01

Mechanical Design

Designed and 3D printed custom delta robot components optimized for paper handling

02

Suction System

Developed a vacuum-based end effector with pressure feedback for delicate paper manipulation

03

Control System

Implemented inverse kinematics algorithms on Arduino for precise positioning

04

Software Interface

Created a Python-based control interface for path planning and operation

Key Features

Precision Control

Achieves sub-millimeter positioning accuracy for detailed paper assembly tasks

Adaptive Suction

Variable suction strength adjusts automatically to different paper weights and textures

High Speed Operation

Capable of performing up to 60 pick-and-place operations per minute

Applications & Impact

SPAC-R demonstrates significant potential for automating paper-based crafting and assembly tasks in educational settings, artistic applications, and small-scale manufacturing. The project contributes to the field of soft robotics by showcasing effective techniques for handling delicate materials with precision and care, while maintaining the speed advantages of delta robot architectures.

Inverted Pendulum: Control and State Estimation

MTech, 2nd semester IIT Gandhinagar Control Systems

Key Innovation

A dual-mode control strategy combining energy-based swing-up control with LQR stabilization, enhanced with state observers for robust performance in noisy environments.

This project presents a control strategy for an inverted pendulum on a cart, tackling the challenge of its inherent instability. The approach employs a dual-mode control system, combining an energy-based swing-up controller to bring the pendulum upright and a Linear Quadratic Regulator (LQR) for fine stabilization once it's near the desired vertical position.

View Project Report

Simulation Demo

Inverted Pendulum Balancing Animation

Methodology

The project involved developing a comprehensive control and state estimation strategy for the inverted pendulum system, addressing both the control challenges and real-world sensor limitations.

01

System Modeling

Derived nonlinear equations using Lagrangian mechanics and linearized around the upright position

02

Dual-Mode Control

Implemented energy-based swing-up control and LQR stabilization with smooth transition

03

State Observers

Designed Luenberger observer and Kalman filter for state estimation from limited measurements

04

Performance Analysis

Conducted comparative analysis of observer performance under varying noise conditions

Key Findings

LQR Performance

LQR gain matrix [185.16, 219.10, -7.81, 6.61] achieved optimal stabilization with minimal control effort

Energy Control

Energy-based swing-up successfully brought pendulum from downward to near-upright position

Kalman Superiority

Kalman filter consistently outperformed Luenberger observer with up to 40% lower RMS error in noisy conditions

Applications & Impact

This research demonstrates a robust approach to controlling inherently unstable systems with limited sensor information, applicable to various domains including robotics, aerospace, and industrial automation. The findings on observer-based control provide valuable insights for implementing control systems in real-world noisy environments, potentially improving the reliability and performance of unstable mechanical systems.

Limb Impedance in Human Drilling and Screwing Tasks

MTech, 2nd semester IIT Gandhinagar Human-Robot Interaction

Key Innovation

Quantitative analysis of how human arm impedance adapts during drilling and screwing tasks under known versus unknown force perturbations, providing insights for human-robot collaborative systems.

This academic project explores how human arm impedance adapts during drilling and screwing tasks when subjected to external force perturbations, comparing responses under known versus unknown perturbation conditions. By analyzing position and force data through a 2D second-order dynamic model, the study estimates mass, damping, and stiffness matrices for the arm.

View Project Report

Drilling Task

Screwing Task

Project Demo

Watch Limb Impedance Demo on YouTube

Methodology

The project involved a carefully designed experimental setup to measure and analyze human arm impedance during common manual tasks under different perturbation conditions.

01

Experimental Setup

Subject's wrist fixed to table with five strings attached to hand, routed through pulleys with weights

02

Task Performance

Subject performed drilling and screwing tasks while maintaining tool position despite perturbations

03

Data Collection

Vicon motion capture system recorded 3D position data of wrist and pulleys across 21 trials

04

Impedance Estimation

Arm modeled as 2D second-order system with least squares regression to estimate parameters

Key Findings

Stiffness Adaptation

Known force conditions resulted in higher, more targeted stiffness along the primary task axis (1.3x higher for drilling, 24% higher for screwing)

Damping Differences

Dramatic differences in damping, with known force condition showing up to 152x higher damping in the primary screwing axis

Stiffness Ellipses

Human arm exhibited natural asymmetry in stiffness, with higher stiffness in the dominant arm

Tension Distribution

Eta vector analysis revealed regions where all cables maintain positive tension for stable control

Applications & Impact

This research has significant implications for rehabilitation robotics, where cable-driven systems can provide controlled assistance over larger and more stable workspaces than human capabilities alone. The findings suggest that cable-driven manipulators with higher redundancy offer enhanced control precision and stability, potentially benefiting rehabilitation practices. However, careful regulation of motion range remains crucial for safe human-robot interaction in practical applications.

Robot-Assisted Human Limb Workspace Analysis

MTech, 2nd semester IIT Gandhinagar Robotics

Key Innovation

Comparative analysis of cable-driven manipulators with human arm workspace, demonstrating enhanced reach capabilities and stability with increased cable redundancy for rehabilitation applications.

This project analyzes the workspace of a two-link planar manipulator actuated by cables, modeling an upper-extremity human limb. The study evaluates the reachable zone of the limb's end-point under different cable redundancy conditions and compares it with anatomically constrained human arm workspace.

View Code (m=n+1) View Code (m=n+2) View Project Report

Workspace with 3 Cables (m=n+1)

Workspace with 4 Cables (m=n+2)

Methodology

The project involved a systematic approach to modeling and analyzing cable-driven manipulators in comparison with human arm workspace limitations.

01

Mathematical Formulation

Derived kinematic equations for the 2-link manipulator and defined cable attachment points

02

Structural Matrix

Formulated generalized structural matrices for 3-cable (m=n+1) and 4-cable (m=n+2) systems

03

Eta Vector Computation

Calculated null space vectors to identify feasible joint configurations where cables maintain tension

04

Workspace Mapping

Mapped feasible joint configurations to task space and compared with human arm model

Key Findings

Enhanced Workspace

Cable-driven manipulators demonstrated broader range of motion than anatomically constrained human arm

Redundancy Benefits

4-cable system (m=n+2) showed improved stability and more uniform workspace distribution

Workspace Asymmetry

Human arm workspace exhibited natural asymmetry due to physiological constraints

Tension Distribution

Eta vector analysis revealed regions where all cables maintain positive tension for stable control

Applications & Impact

This research has significant implications for rehabilitation robotics, where cable-driven systems can provide controlled assistance over larger and more stable workspaces than human capabilities alone. The findings suggest that cable-driven manipulators with higher redundancy offer enhanced control precision and stability, potentially benefiting rehabilitation practices. However, careful regulation of motion range remains crucial for safe human-robot interaction in practical applications.

Overhead Press Simulation and Analysis

MTech, 2nd semester IIT Gandhinagar Human-Robot Interaction

Key Innovation

Comprehensive biomechanical analysis of the overhead press exercise using musculoskeletal modeling to estimate joint kinematics, muscle forces, and joint torques for applications in rehabilitation, sports science, and human-robot interaction.

This project focuses on simulating and analyzing the biomechanics of a human overhead press exercise using musculoskeletal modeling tools. Understanding human movement biomechanics is crucial for various fields, including rehabilitation, sports science, ergonomics, and human-robot interaction. The study utilized OpenSim, a powerful software tool for simulating, analyzing, and visualizing internal dynamics such as joint kinematics, muscle forces, and joint torques, which are typically difficult to measure non-invasively.

View Project Report

Methodology

The project employed a comprehensive approach to analyze the biomechanics of the overhead press using motion capture data and musculoskeletal modeling.

01

Data Collection

Captured motion data from subjects performing overhead press exercises with varying loads

02

Model Scaling

Adapted generic musculoskeletal models to match subject-specific anthropometry

03

Inverse Kinematics

Calculated joint angles throughout the movement trajectory

04

Inverse Dynamics

Computed joint moments and forces required to produce the observed motion

05

Muscle Analysis

Estimated individual muscle contributions and activation patterns

Key Findings

Load Distribution

Deltoid and trapezius muscles showed highest activation during the concentric phase

Joint Coordination

Synchronized shoulder and elbow movement patterns optimized force generation

Compensatory Mechanisms

Increased trunk extension observed with heavier loads to maintain stability

Energy Transfer

Efficient energy transfer between joints reduced overall metabolic cost

Applications & Impact

This research contributes valuable insights into the biomechanics of overhead pressing movements, with applications in rehabilitation protocol design, sports performance optimization, and the development of assistive robotic devices. The detailed muscle force estimations and joint kinetics analysis provide a foundation for understanding injury mechanisms, optimizing training techniques, and designing more effective human-robot interaction systems that complement natural human movement patterns.

Cartesian Impedance Control on HEAL

Cartesian Impedance Control for Robot Manipulation

Research Project IIT Gandhinagar Human-Robot Interaction

Key Innovation

Implementation of a Cartesian impedance controller that enables compliant human-robot collaborative manipulation, allowing a robot to adaptively assist humans in lifting and placing tasks.

This project implements a Cartesian Impedance Controller for a robotic manipulator collaboratively lifting and placing a box with a human partner. The robot operates in task space (Cartesian space) and exhibits compliant, human-friendly behavior.

View on GitHub

Project Demo

Watch Cartesian Impedance Control Demo on YouTube

Core Concept & Implementation

The end-effector is modeled as a mass-spring-damper system in 3D space, allowing the robot to follow smooth position and orientation trajectories while adapting compliantly to external forces — ideal for human-robot interaction tasks.

01

Kinematics Handling (via KDL)

Computes forward kinematics for pose tracking, Jacobian for mapping Cartesian forces to joint torques, and updates pose and velocity in real-time

02

Trajectory Generation

Uses quintic polynomial interpolation for position and SLERP for orientation, ensuring smooth acceleration/deceleration with zero start/end velocity

03

Execution Flow

Sequences through initialize, lift, and place poses with a 15-second interpolation per pose, maintaining final position with impedance control

04

Control Loop

Runs at 500 Hz, continuously computing Cartesian errors and applying impedance torques via Jacobian transpose

Key Features

Human-Robot Collaboration

Enables safe and compliant co-manipulation with a human lifting the other side of the box

Smooth Trajectories

Quintic polynomial interpolation ensures jerk-free motion with p(t) = a₀ + a₁t + a₂t² + a₃t³ + a₄t⁴ + a₅t⁵

Adaptive Response

Automatically adjusts to external disturbances or human corrections during the task

Safety Features

Avoids excessive joint forces through natural impedance-based compliance

Applications & Impact

This Cartesian impedance control strategy demonstrates significant potential for collaborative robotics applications, particularly in manufacturing, logistics, and assistive scenarios. The controller's ability to maintain precise positioning while adapting to human input makes it ideal for co-manipulation tasks, physical human-robot interaction (pHRI), and assistive lifting operations. The implementation provides a foundation for developing more sophisticated collaborative behaviors that can enhance productivity and safety in human-robot workspaces.

Joint Impedance Control on HEAL

Joint Impedance Control for Robot Manipulation

Research Project IIT Gandhinagar Robot Control

Key Innovation

Implementation of a joint-space impedance controller with velocity-based trajectory planning that ensures compliant, smooth, and velocity-safe motion for robotic manipulation tasks.

This project implements joint-space impedance control with PID gains and quintic trajectory generation for the Heal Robot. The controller computes joint torques based on position, velocity, and integral errors, ensuring compliant and smooth motion. The implementation supports effort (torque) control, is designed for ROS, and integrates trajectory planning based on maximum time and velocity.

View on GitHub

Core Components & Implementation

The controller uses a PID torque control structure enhanced with impedance behavior, combined with velocity-based trajectory planning for safe and efficient motion.

01

PID Control Parameters

Carefully tuned gains (Kp: [148, 140, 130, 120, 115, 130], Kd: [57, 52, 45, 50, 40, 45], Ki: [65, 60, 60, 60, 55, 60]) for optimal performance across all joints

02

Control Law

Computes torque as τ = Kp × e_pos + Kd × e_vel + Ki × ∫ e_pos dt with integral clamping to prevent windup

03

Velocity-Based Planning

Calculates trajectory duration based on joint distances and maximum allowable velocities, ensuring no joint exceeds its speed limit

04

Quintic Trajectory

Generates smooth paths using 5th-order polynomials: q(t) = a₀ + a₁t + a₂t² + a₃t³ + a₄t⁴ + a₅t⁵ with zero initial/final velocities and accelerations

Key Features

Dynamic Duration

Computes trajectory time based on per-joint distance and maximum velocity constraints

Smooth Motion

Quintic polynomial ensures jerk-free transitions with continuous acceleration profiles

Safety Features

Enforces velocity limits and implements anti-windup strategy for integral term

High Responsiveness

Control loop runs at 500 Hz with target-aware tracking and termination on convergence

Applications & Impact

This joint impedance control implementation provides a robust foundation for precise and compliant robot manipulation tasks. The velocity-based trajectory planning approach ensures safe operation while maintaining optimal performance. The system is particularly valuable for applications requiring careful motion control, such as assembly tasks, human-robot collaboration, and manipulation in constrained environments. The controller's ability to balance between position accuracy and compliance makes it suitable for a wide range of robotic applications.

Admittance Control on HEAL

Admittance Control on 6-DOF Heal Robot

Research Project IIT Gandhinagar Robotics

Key Innovation

Implementation of an admittance control strategy using force-torque sensor feedback to enable intuitive physical human-robot interaction, allowing the robot to respond naturally to applied forces.

This project implements admittance control on a custom 6-DOF Heal robot with a biaxial FUTEK force-torque (FT) sensor mounted on the end-effector. The sensor measures forces in the X and Y directions, which are then used to command motion through an admittance control strategy in Cartesian space.

View on GitHub

Force-Torque Sensor Mounted on HEAL Robot's End-Effector

Project Demo

Admittance Control in XY Plane Demonstration

System Dynamics & Implementation

The admittance control strategy maps measured forces to robot motion using a quasi-static approximation of the system dynamics.

01

Dynamic Equation

Based on the fundamental equation: M·Δẍ + Kd·Δẋ + he = h, where M is the mass matrix, Kd is the damping matrix, he is the end-effector wrench, and h is the desired wrench

02

Quasi-Static Approximation

Neglects inertial effects to derive the control law: ẋr = ẋd + Kd⁻¹(he - h), mapping measured forces to reference velocities

03

Jacobian Mapping

Uses the robot Jacobian to convert Cartesian velocities to joint velocities: q̇r = q̇d + J⁻¹[Kd⁻¹(he - h)]

04

Sensor Integration

Processes real-time force measurements from the biaxial FUTEK FT sensor to enable responsive control in both X and Y directions

Key Features

Multi-Directional Control

Supports both 1D (Y-axis only) and 2D (X-Y plane) admittance control modes

Force-to-Velocity Mapping

Direct mapping of measured forces to Cartesian velocities with tunable damping parameters

Intuitive Interaction

Enables natural physical guidance of the robot through direct hand contact

Adjustable Compliance

Tunable damping values allow for customizing the robot's responsiveness to applied forces

Applications & Impact

This admittance control implementation enables intuitive physical human-robot interaction, making it valuable for collaborative tasks, teaching by demonstration, and assistive applications. The system allows users to guide the robot naturally by applying forces directly to the end-effector, creating a more accessible interface for non-expert users. This approach has significant potential in manufacturing, healthcare, and service robotics where direct physical interaction between humans and robots is beneficial.

Object Admittance Control for Dual-Arm Robot

Object Admittance Control for Dual-Arm Robot in MuJoCo

Research Project IIT Gandhinagar Advanced Robotics Control

Key Innovation

Implementation of object-level admittance control for coordinated dual-arm manipulation, enabling compliant and adaptive object handling in response to external forces.

This project implements object admittance control for a dual-arm robotic system in a MuJoCo simulation. The robot collaboratively lifts an object (like a box) using sensor feedback, dynamic control, and intelligent coordination between both arms.

View on GitHub

Project Demo

Watch Object Admittance Control Demo on YouTube

Core Concepts & Implementation

The object admittance control strategy enables compliant interaction with the environment by responding to external forces detected through both robot arms.

01

Admittance Control Equation

Based on the equation: ẋₒ* = ẋₒ + D⁻¹(W - W* - K(xₒ* - xₒ)), where D is the damping matrix, K is the stiffness matrix, and W represents the sensed wrench

02

Contact Wrench Detection

Detects contact between robot and box, extracting forces & torques (wrenches) from both arms for feedback control

03

Velocity Mapping

Converts Cartesian object velocity to joint velocity using Jacobians of both arms with safety velocity limits

04

Control State Machine

Manages transitions between reaching, grasping, lifting, and maintaining states based on robot progress and sensor feedback

Key Features

Dual-Arm Coordination

Combines control of two robot arms to stably grasp and move objects with coordinated motion

Force-Based Lifting

Adjusts lifting motion based on sensed wrenches from both arms for adaptive object handling

Fallback Logic

Implements intelligent recovery strategies when contact is lost or force thresholds are not met

MuJoCo Simulation

Leverages accurate physics-based simulation for realistic testing of control algorithms

Applications & Impact

This object admittance control implementation demonstrates significant potential for collaborative robotic manipulation tasks. By enabling compliant and adaptive object handling, the system can be applied to various scenarios including manufacturing assembly, logistics, healthcare assistance, and household tasks. The dual-arm coordination with force feedback creates more robust manipulation capabilities that can adapt to object variations and environmental uncertainties, advancing the field of physical human-robot interaction and collaborative robotics.

Impedance Control on 2R Manipulator

Impedance Control Implementation on 2R Manipulator

Research Project IIT Gandhinagar Robot Control

Key Innovation

Implementation of progressive control strategies from basic gravity compensation to full impedance control on a 2R manipulator, enabling compliant and adaptive interaction with the environment.

In our 2R manipulator simulation using MuJoCo, we implemented and explored various control strategies to achieve stable and responsive joint behavior. The project demonstrates the evolution of control approaches from basic compensation to interactive impedance control.

View on GitHub

Control Strategies

Our research explored a progression of control approaches, starting with basic gravity compensation and advancing toward full impedance control for interactive tasks.

01

Gravity Compensation

Calculated and applied torques at each joint to counteract gravity effects, enabling the manipulator to maintain static configurations without collapsing

02

PD Control

Implemented Proportional-Derivative control with desired joint positions, computing control torques based on position and velocity errors

03

Impedance Control

Combined gravity compensation and PD control to emulate a mass-spring-damper system, allowing compliant response to external forces

04

Simulation Testing

Validated control strategies in MuJoCo environment with various initial conditions and external disturbances

Key Findings

Static Stability

Gravity compensation successfully maintained static configurations (e.g., 45° and 60° for joints 1 and 2) without collapsing

Dynamic Response

PD control enabled smooth convergence to target configurations from arbitrary initial states

Compliance Benefits

Impedance control framework demonstrated potential for human-robot interaction and adaptive object manipulation

Framework Extensibility

The simulation provides a foundation for integrating advanced behaviors like surface contact maintenance and constrained path following

Applications & Future Work

This research establishes a foundation for more advanced robot control applications. In future extensions, impedance control will allow the 2R arm to maintain contact with surfaces, guide objects along constrained paths, or adaptively respond to perturbations rather than rigidly fighting against them. The framework developed here serves as a stepping stone toward more sophisticated human-robot interaction scenarios and compliant manipulation tasks.

About Me

My Achievement Journey

Academic & Professional Milestones

Key Achievement

Achievement Timeline

Best Student Award (2015)

National Player in throwball (2018)

Jabalpur Engineering College (2020)

IIT Indore Exchange (2023)

Gold Medalist (2024)

IIT Gandhinagar Research (2024-Present)

Future Directions

Projects

BTech Projects

Analyzing the Impact of Increasing Leakage Fault in Single and Multiple Cylinders and Pressure Fluctuation Reduction via Piston Rearrangement in an Axial Piston Pump

Key Innovation

Methodology

Simulation Development

Fault Analysis

Validation

Piston Arrangement Study

Key Findings

Fault Progression

Performance Impact

Optimal Configuration

Conclusion & Impact

SPAC-R: Suction-Driven Paper Assembly and Crafting Robot

SPAC-R: Suction-Driven Paper Assembly and Crafting Robot

Key Innovation

Project Demo

Methodology

Mechanical Design

Suction System

Control System

Software Interface

Key Features

Precision Control

Adaptive Suction

High Speed Operation

Applications & Impact

Inverted Pendulum: Control and State Estimation

Inverted Pendulum: Control and State Estimation

Key Innovation

Simulation Demo

Methodology

System Modeling

Dual-Mode Control

State Observers

Performance Analysis

Key Findings

LQR Performance

Energy Control

Kalman Superiority

Applications & Impact

Limb Impedance in Human Drilling and Screwing Tasks

Limb Impedance in Human Drilling and Screwing Tasks

Key Innovation

Project Demo

Methodology

Experimental Setup

Task Performance

Data Collection

Impedance Estimation

Key Findings

Stiffness Adaptation

Damping Differences

Stiffness Ellipses

Tension Distribution

Applications & Impact

Robot-Assisted Human Limb Workspace Analysis

Robot-Assisted Human Limb Workspace Analysis

Key Innovation

Methodology

Mathematical Formulation

Structural Matrix

Eta Vector Computation

Workspace Mapping

Key Findings

Enhanced Workspace