Haptic Interface Virtual Worlds

Completed Last Updated: 2025-02-28

Quick Navigation

• Project Overview
• Implementation
• Virtual Worlds
• Results & Analysis

Project Overview

This project explores the design and implementation of virtual haptic environments using an S32K144 microcontroller. By combining quadrature decoding for position sensing with Pulse Width Modulation (PWM) for motor control, I created a system that can simulate various physical properties like springs, walls, and dampers through touch.

The haptic interface provides force feedback based on the position of a rotary wheel, allowing users to physically “feel” virtual objects through the sense of touch. This project demonstrates fundamental concepts in embedded control systems, digital signal processing, and human-computer interaction.

Implementation

The system architecture combines multiple embedded control techniques:

Hardware Components

• S32K144 Microcontroller - Core processing unit
• DC Motor with Encoder - Provides haptic feedback and position sensing
• Switched Capacitor Filter - 4th order Butterworth filter for PWM-to-analog conversion
• Power Amplifier - Drives the motor with the appropriate current

Key Technical Features

• Quadrature Decoding - Tracks wheel position with 4000 counts per revolution
• PWM Modulation - Generates precise duty cycles (24%-76%) at 20kHz
• Real-time Control Loop - Reads position, calculates forces, and updates PWM output
• Torque Mapping - Converts desired torque (N·mm) to appropriate duty cycle values

Hardware architecture showing the signal flow from microcontroller to haptic feedback

Virtual Worlds

I implemented several virtual haptic environments, each with unique physical properties:

Virtual Spring

The spring model applies a restoring torque proportional to angular displacement:

float virtualSpring(float angle)
{
    float k_spring = 5;  // Spring constant (N·mm/degree)
    float torque = -k_spring * angle;
    return torque;
}

This creates a sensation of pushing against a spring, with increasing resistance as the wheel rotates further from the center position.

Virtual Wall

The wall model creates a rigid boundary at a specific angular position:

float virtualWall(float angle)
{
    float torque = 0;
    float k_wall = 500;  // Wall stiffness (N·mm/degree)
    if (angle < WALL_POSITION)
    {
        torque = -k_wall * (angle - WALL_POSITION);
    }
    return torque;
}

Users feel free movement on one side of the boundary, but encounter resistance when attempting to move through the virtual wall.

Visualization of virtual spring (left) and virtual wall (right) force profiles

Results & Analysis

The implementation demonstrated several interesting findings:

Spring Oscillation Analysis

For a spring constant of 5 N·mm/degree with a wheel moment of inertia of 6.4×10^-4 kg·m²/radian, the theoretical oscillation frequency was 1.4 Hz. In practice, the measured oscillation frequency was 1.2 Hz, with the difference attributed to:

• Software execution time creating a delay in the control loop
• Mechanical friction providing damping to the system
• Motor dynamics affecting the response

Wall Stiffness Trade-offs

While increasing the wall spring constant (k_wall) made the wall feel more rigid, extremely high values (>1000 N·mm/degree) resulted in oscillatory behavior as the system repeatedly corrected position. The optimal value was found to be approximately 500 N·mm/degree, providing a good balance between:

• Wall rigidity and stability
• Smooth transitions between free movement and wall contact
• Prevention of torque saturation (limited to ±800 N·mm)

The project successfully demonstrated that convincing haptic environments can be created with relatively simple control algorithms and careful tuning of system parameters.

×