⚡ electrical engineering

Co-Simulation and Comparative Control Analysis of a 3-DOF Serial Manipulator Using MSC ADAMS and MATLAB

This paper presents a co-simulation framework using MSC ADAMS and MATLAB to model a 3-DOF serial manipulator and demonstrates that an adaptive Fuzzy-PID controller significantly outperforms a conventional PID controller by drastically reducing tracking error, overshoot, and settling time while maintaining high positioning accuracy.

Original authors: Mohammad Heidar Khamsehei Fadaei, Seyed Ali Hamzeh Pahnehkolaei

Published 2026-06-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Mohammad Heidar Khamsehei Fadaei, Seyed Ali Hamzeh Pahnehkolaei

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to teach a robotic arm to pick up a cup and move it to a saucer without spilling a drop. To do this, you need two things: a perfect understanding of how the arm's body moves (the physics) and a smart brain to tell the motors exactly how hard to push (the control).

This paper is about building a virtual test lab to figure out the best way to teach that robotic arm, and then proving that a "smart" brain works much better than a "dumb" one.

Here is the breakdown of their work, using simple analogies:

1. Building the Virtual Robot (The "Digital Twin")

Instead of building a real metal robot and risking breaking it, the researchers used a computer program called MSC ADAMS. Think of this as a high-end video game engine that doesn't just look real, but feels real.

The Build: They created a 3-jointed robotic arm (like a human shoulder, elbow, and wrist) entirely in the computer.
The Physics: They gave every part of the robot the exact weight and stiffness of real steel. They calculated how heavy each part is and how it swings around. This ensures that when the computer simulates a movement, it behaves exactly like a real machine would, including all the tricky physics like momentum and inertia.

2. The "Brain" of the Operation (Co-Simulation)

The researchers connected their virtual robot (in ADAMS) to a control program (MATLAB/Simulink).

The Analogy: Imagine the ADAMS program is the body of the robot, and the MATLAB program is the brain. They are wired together so the brain can send commands ("Move your elbow!") and the body instantly reports back ("I'm moving, but I'm heavy and slowing down").
Why do this? This allows them to test different "brains" (control strategies) on the same "body" without ever needing to build a physical prototype. It saves money and time.

3. The Two Brains: "The Stickler" vs. "The Adaptive Coach"

The researchers tested two different ways to control the robot's movement:

Controller A: The Standard PID (The "Stickler")
- How it works: This is the traditional method used in factories. It's like a strict teacher who gives the same amount of help no matter what. If the robot is off-target, the teacher pushes a little harder. If it's way off, the teacher pushes a lot harder.
- The Problem: The robot's body changes as it moves. When the arm swings fast, it creates its own forces (like centrifugal force). A "Stickler" doesn't know how to adjust for these changes, so it often overshoots (goes too far) or wobbles before stopping.
Controller B: The Fuzzy-PID (The "Adaptive Coach")
- How it works: This is a smarter brain that uses "Fuzzy Logic." Think of it as a coach who watches the robot in real-time. If the robot is moving fast and heavy, the coach says, "Okay, I need to be gentler." If the robot is struggling to start, the coach says, "Push harder!"
- The Magic: It constantly tweaks its own settings (gains) based on what is happening right now. It's not following a rigid rulebook; it's adapting to the situation.

4. The Results: Who Won the Race?

They asked both "brains" to guide the robot along a complex path. Here is how the "Adaptive Coach" (Fuzzy-PID) beat the "Stickler" (Standard PID):

Accuracy: The "Stickler" missed the target by a noticeable margin. The "Adaptive Coach" was incredibly precise, staying within 0.25 millimeters of the target (about the thickness of a human hair).
Overshooting: The "Stickler" often swung past the target and had to swing back (like a pendulum). The "Adaptive Coach" stopped almost exactly where it was supposed to.
- The Paper's Claim: The new controller reduced the "overshoot" (swinging too far) by 78%.
Speed to Stop: When the robot needed to stop, the "Stickler" took a long time to settle down and stop wobbling. The "Adaptive Coach" settled down much faster.
- The Paper's Claim: It reduced the time to settle by 64%.
Error Reduction: Overall, the tracking error (how far off the path the robot went) was cut by 67%.

5. The Big Takeaway

The paper concludes that using this Virtual Prototyping method is a game-changer.

No Wasted Metal: You don't need to build, break, and rebuild physical robots to test ideas. You can do it all in the computer.
Better Control: For complex machines like robotic arms, a "smart" controller that adapts in real-time is far superior to a standard, fixed controller.
Cost & Time: The authors claim this approach could cut development costs by 40-60% and turn design cycles that used to take weeks into something that takes days.

In short: They built a perfect digital robot, gave it a "smart" brain that learns as it moves, and proved that this combination is much faster, more accurate, and more stable than using a traditional, rigid controller.

Technical Summary: Co-Simulation and Comparative Control Analysis of a 3-DOF Serial Manipulator

Problem Statement
The effective deployment of serial manipulators in industrial robotics requires rigorous modeling of kinematic and dynamic behaviors coupled with sophisticated control strategies to ensure accurate trajectory tracking and disturbance rejection. However, the inherent nonlinearities, inertial coupling, and velocity-dependent Coriolis and centrifugal forces in multi-degree-of-freedom (DOF) systems present significant challenges for controller design. Conventional Proportional-Integral-Derivative (PID) controllers, while ubiquitous due to their simplicity, often rely on fixed gains that struggle to adapt to varying operating conditions and dynamic loads. Furthermore, the development of robust control systems traditionally relies on physical prototyping, which is costly and time-consuming. This study addresses the need for a high-fidelity virtual prototyping environment and an adaptive control strategy capable of managing the complex dynamics of a planar 3-DOF serial manipulator.

Methodology
The research employs a comprehensive co-simulation framework integrating MSC ADAMS for multi-body dynamic modeling and MATLAB/Simulink for control system design. The methodology proceeds through four primary stages:

Mechanical Modeling: A planar 3-DOF serial manipulator consisting of six rigid links and six revolute joints was modeled in MSC ADAMS. Components were assigned steel material properties (density: $7.801 \times 10^{-6}$ kg/mm³, Young's Modulus: $2.07 \times 10^5$ N/mm²). Inertial parameters, including mass and moments of inertia, were automatically calculated by ADAMS based on geometric volumes.
Kinematic and Dynamic Formulation:
- Kinematics: The Denavit-Hartenberg (D-H) convention was used to derive forward and inverse kinematic equations, mapping joint displacements to end-effector pose and vice versa.
- Dynamics: The Euler-Lagrange formulation was utilized to derive the equations of motion, explicitly accounting for the symmetric positive-definite inertia matrix and nonlinear terms representing Coriolis, centrifugal, and gravitational effects.
Co-Simulation Framework: The ADAMS mechanical model was exported to MATLAB/Simulink via the ADAMS/Control interface. This created a unified environment where Simulink implemented control algorithms (running at 1 kHz) while ADAMS resolved the physical dynamics (at 100 μs intervals). The interface exchanged joint torque inputs and joint position/velocity outputs.
Control Strategy Design: Two controllers were designed and compared:
- Conventional PID: A fixed-gain PID controller tuned for each joint.
- Adaptive Fuzzy-PID: A hybrid controller employing a fuzzy inference system to dynamically schedule PID gains ( $K_p, K_i, K_d$ ) in real-time based on the tracking error ( $e$ ) and its derivative ( $\Delta e$ ). The fuzzy logic system utilized seven Gaussian membership functions for inputs and singleton/integer outputs for gain adjustments.

Key Contributions

Integrated Virtual Prototyping: The study establishes a robust co-simulation workflow that validates theoretical kinematic and dynamic models against high-fidelity multi-body simulations, demonstrating the viability of virtual prototyping for reducing development costs and accelerating design cycles.
Adaptive Control Implementation: The paper details the design and implementation of a Fuzzy-PID controller specifically tailored for the nonlinear dynamics of a 3-DOF manipulator, utilizing real-time gain scheduling to compensate for inertial coupling and varying loads.
Comparative Performance Analysis: A rigorous quantitative evaluation is provided, comparing the fixed-gain PID and adaptive Fuzzy-PID controllers across multiple performance metrics, including joint tracking error, overshoot, settling time, and end-effector positioning accuracy.

Results
Simulation results indicate that the adaptive Fuzzy-PID controller significantly outperforms the conventional PID controller:

Tracking Accuracy: The Fuzzy-PID controller achieved a 67% reduction in joint tracking error compared to the PID controller. Specifically, the Root Mean Square Error (RMSE) for Joint 1 dropped from 0.36° to 0.12°.
Transient Response: The Fuzzy-PID approach reduced overshoot by 78% on average. For Joint 1, overshoot decreased from 37.3% (PID) to 8.2%. Joint 3 saw the elimination of overshoot entirely (0% vs. 1.4%).
Settling Time: An average improvement of 64% in settling time was observed across all joints. For example, Joint 1's settling time improved from 0.275 s to 0.098 s.
End-Effector Precision: The Fuzzy-PID controller maintained end-effector positioning accuracy within ±0.25 mm throughout complex trajectories, whereas the PID controller exhibited deviations up to ±0.82 mm.

Significance and Claims
The paper claims that the proposed methodology validates the theoretical models derived via D-H and Lagrangian methods, as evidenced by the close correlation between analytical predictions and simulation results. The study underscores the significant advantages of virtual prototyping, estimating a 40–60% reduction in development costs and a shortening of design iteration cycles from weeks to days.

The authors assert that the adaptive Fuzzy-PID controller provides superior robustness and performance for nonlinear robotic systems by effectively compensating for dynamic coupling and inertial variations that challenge fixed-gain controllers. The research concludes that this co-simulation framework offers a practical, cost-effective foundation for the design and optimization of industrial robotic systems, with future work suggested for spatial manipulators and the integration of advanced factors like joint flexibility and external disturbances.