OttolinkPCB

Design and Control of a 4-DoF Anthropomorphic Robotic Manipulator

Ottolink is a custom 4-DoF anthropomorphic robotic manipulator designed and built for spatial end-effector trajectory execution. The project combines mechanical design, embedded real-time control, Jacobian-based inverse kinematics, custom electronics, and MuJoCo simulation into a complete robotic system.

Overview

Ottolink features a 3-DoF shoulder configuration and 1-DoF pitch elbow, providing spatial control of the end effector.

The project was developed as a complete robotics pipeline:

CAD → Mechanical Assembly → Kinematics → Simulation → Embedded Control → Custom PCB → Physical Robot

The controller runs on an ATmega328P-based embedded system and executes Cartesian-space trajectories using a Jacobian pseudoinverse controller.

Watch It In Action

You can watch the arm in action on my Portfolio Website.

Features

• 4-DoF anthropomorphic robotic arm
• 3-DoF shoulder + 1-DoF pitch elbow
• Custom mechanical design in Autodesk Inventor
• 3D-printable mechanical components
• Jacobian pseudoinverse inverse kinematics
• Real-time Cartesian-space control
• Custom C++ linear algebra library
• Rodrigues rotation formulation for joint-axis transformations
• Multi-axis synchronized PWM servo control
• Joint interpolation and trajectory execution
• Custom control PCB designed in KiCad
• Dedicated power distribution and voltage regulation
• MuJoCo simulation environment
• Simulation-to-hardware trajectory validation

Mechanical Design

The robotic arm was designed and assembled in Autodesk Inventor, with component geometry optimized for 3D printing tolerances.

The mechanical design considers:

• Joint clearances
• Structural rigidity
• Servo mounting
• Payload-to-weight considerations
• Assembly constraints
• Joint-axis alignment

The final design was manufactured using 3D-printed components and integrated with the actuator and control system.

CAD

The CAD/ directory contains the mechanical design files and supporting models.

Kinematics and Control

Jacobian-Based Cartesian Control

Ottolink uses a Jacobian pseudoinverse controller to map Cartesian end-effector errors into joint-space corrections.

Given the end-effector error:

e = x_target - x_current

the joint-space update is calculated using:

Δq = J⁺ e

where:

• J is the manipulator Jacobian
• J⁺ is the Moore-Penrose pseudoinverse
• e is the Cartesian position error
• Δq is the resulting joint-space correction

The controller iteratively updates the joint configuration to reduce the Cartesian position error and converge toward the target trajectory.

Rodrigues Rotation

Rodrigues' rotation formula is used to represent arbitrary rotations around the robot's joint axes.

This is particularly important for handling the physical robot's non-ideal axis configurations and maintaining consistency between:

• The CAD model
• The mathematical kinematic model
• The MuJoCo simulation
• The physical robot

Custom Linear Algebra Library

A lightweight custom C++ matrix and linear algebra library was developed specifically for the ATmega328P.

The library provides the operations required by the real-time controller, including:

• Matrix multiplication
• Matrix transposition
• Matrix inversion
• Pseudoinverse operations
• Vector operations

The implementation was designed around the memory and computational constraints of the ATmega328P rather than relying on a large external numerical library.

Embedded Controller

The physical robot is controlled by an ATmega328P-based embedded system.

The firmware provides:

• Multi-axis PWM control
• Synchronized servo actuation
• Joint interpolation
• Cartesian trajectory execution
• Real-time kinematic calculations
• Joint-space corrections

The controller was implemented using C++ and PlatformIO.

Firmware

The embedded controller can be found in:

OttolinkController/

The main source code is located in:

OttolinkController/src/

Electronics

A custom control PCB was designed in KiCad.

The PCB provides:

• Servo power distribution
• Localized voltage regulation
• High-current actuator power handling
• Separation of actuator power from logic-level control signals
• Centralized connections between the microcontroller and actuators

The board was designed to reduce wiring complexity while providing a more robust electrical interface for the robotic system.

PCB Design

PCB source files are available in:

OttolinkPCB/

The directory contains the KiCad schematic, PCB layout, project files, and custom footprint data required for the board.

Manufacturing files are available in:

PCBGerberFiles/

Simulation

A complete MuJoCo simulation of the robotic arm was developed to validate the kinematic and control system before deployment to the physical hardware.

The simulation includes:

• Robot geometry
• Coordinate frames
• Joint definitions
• Joint limits
• Kinematic transformations
• End-effector control
• Cartesian trajectory execution

The simulation files are located in:

Sim/
├── meshes/
├── ottolink.xml
└── OttolinkSimulation.py

The simulation was used to validate:

1. Forward kinematics
2. Coordinate-frame transformations
3. Jacobian calculations
4. Jacobian pseudoinverse control
5. Cartesian trajectory generation
6. Controller behavior prior to physical deployment

Simulation-to-Hardware Workflow

The project was developed with a simulation-first workflow.

        Mechanical Design
               │
               ▼
        Kinematic Model
               │
               ▼
         MuJoCo Model
               │
               ▼
      Controller Validation
               │
               ▼
      Embedded Implementation
               │
               ▼
       Physical Robot

This workflow allowed the kinematic and trajectory-control algorithms to be tested in simulation before being executed on the embedded controller and physical hardware.