Rong-Hao Liang

Interactive Intelligent Products

DBB220 / 2019 Semester B Quartile 4 / TU Eindhoven

This course introduces fundamental skills and knowledge for rapid prototyping interactive products with machine learning and signal processing.

Responsible Lecturer: Dr. Rong-Hao Liang (Since 2017)

Assistant Professor, Future Everyday Group, Industrial Design, TU Eindhoven; Assistant Professor, Signal Processing Systems Group, Electrical Engineering, TU Eindhoven



Sensors not only enable interactivity of products but also generate data. Machine learning algorithms leverage computational power and data to empower further the product to deal with design problems involved in prediction, decision, and adaptation. Tangible products support rich embodied interactions. This course aims to help the students understand the main paradigms in sensing, data collection, signal processing, and machine learning to apply them in meaningful design solutions of intelligent and interactive tangible products.

An Interactive Intelligent Product

This course belongs to both Industrial Design competency areas “Math, Data, and Computing” and “Technology and Realization”.

The course will combine theories and practices. Through a series of lectures and workshops, you will learn the principle and functionality of the sensors and machine learning algorithms. You will develop the ability to use these signal processing methods and machine learning algorithms to deal with everyday life problems where real-world complexity, uncertainty, and changing conditions make the use of these technical solutions a necessity. Understanding the type of problems that really will benefit from the utilization of signal processing methods and machine learning algorithms and creating realistic scenarios of use for intelligent products and services is equally important.

The course will cover but is not limited to the following topics:

Course Overview: Topics Covered in Interactive Intelligent Products

Learning Objects

Students will

  1. Understand and be able to apply and evaluate at least one machine learning algorithm correctly.
  2. Understand and be able to use at least one sensor for interactivity and data collection.
  3. Show the ability to use machine learning algorithms as a design method for developing at least one product/service concept in which the machine intelligence augments the design.
  4. Show the ability to address the concept as a realistic design problem in a societal context.

Teaching Assistants

Dr. Zengrong Guo, Postdoc Researcher, Future Everyday Group, Industrial Design, TU Eindhoven

Ruben van Dijk, PhD candidate, Future Everyday Group, Industrial Design, TU Eindhoven


1. Introduction and Problem Formulation

1.1 Introduction

1.2 Problem Formulation

2. Data Preparation and Serial Communication

2.1 Data Preparation

2.2 Serial Communication

3. Classification and Regression

3.1 Classification

3.2 Regression

4. Time-Series Signal Processing

4.1 Time-Series Signal Processing

4.2 **Real-Time Motion Classification and Regression

5. Evaluation and Reporting

5.1 More Algorithms

5.2 Evaluation and Reporting

6. Frequency-Domain Signal Processing

6.1 Extraction Features in Frequency Domain

6.2 Recognizing Sounds and Vibrations

7. Sensor Fusion and Feature Selection

7.1 Sensor Fusion and Context Recognition

7.2 Feature Selection and Dimensionality Reduction

8. Spatial-Domain Signal Processing

8.1 Extracting Features from Images

8.2 Camera-based Activity Recognition

9. Neural Networks and Wrap Up

9.1 Neural Networks

9.2 Looking Back and Moving Forward