Research Project

Autonomous Bicycle

Project Duration

2017–2019

Role and Contribution

Sole Developer & Project Lead (100% Contribution)

Technologies and Skills Applied

Embedded Hardware Design (OrCAD: Schematic Design, PCB Layout)
Embedded Software Development (ATmega, STM32 in C with RTOS)
Digital Instrument Cluster Development (Qt, QML)
Computer Vision Algorithm Development (OpenCV, C++)
Robot Control Algorithm Development (STM32, CMSIS-DSP)
Mechanical Design / CAD (SolidWorks)
Hardware Integration and Electrical Installation

Introduction

During my undergraduate studies, I developed an autonomous bicycle to overcome the inherent instability of two-wheeled vehicles. Progressing from a self-balancing prototype to a full-size model, this two-stage project earned multiple awards and widespread media coverage.

Phase 1 (2017) – Scale Model: Self-Balancing Bicycle

Hardware and System Design

For the first stage of the project, I developed a self-balancing scaled prototype using a Raspberry Pi as the main processing unit. The physical hardware was designed in CAD and manufactured via 3D printing. To drive the system, I engineered a dedicated printed circuit board (PCB)—marking my first time designing a PCB—which fully integrated the motor control IC, battery protection circuit, and sensor interfaces.

Control Algorithms and Autonomy

To achieve self-balancing, an Inertial Measurement Unit (IMU) continuously measures the bicycle’s tilt, while a PD controller calculates the precise steering angle required to maintain equilibrium at a control frequency exceeding 50 Hz. Furthermore, a front-facing MIPI camera utilises OpenCV to detect target destinations, enabling the prototype to navigate autonomously while balancing.

Recognition: In recognition of its innovation, the system received multiple honours at POSTECH’s Spring 2017 Creative IT Design Competition, being awarded both the Grand Prize and the VADAS Award.

Additionally, the project won First Prize at the Creative ICT Convergence Forum, held at COEX in Seoul.

Scale Model of Self-Balancing Bicycle (May 2017)

First Prize – Creative ICT Convergence Forum 2017 (Dec 2017)

Grand Prize – Creative IT Design Competition, Spring 2017 (Jun 2017)

VADAS Award (Special Prize) –
Creative IT Design Competition, Spring 2017 (Jun 2017)

Phase 2 (2018–2019) – Full-Scale Autonomous Bicycle

Introduction

Following the successful feasibility study with the scale model, Phase 2 focused on developing a full-scale autonomous bicycle. To maximise customisability, a standard bicycle was retrofitted with various modules, including a digital instrument cluster, a custom battery pack, and high-torque motors for steering and propulsion.

Implementation

A digital instrument cluster serves as the user interface, developed using Qt (C++) and QML on the NVIDIA Jetson platform. This interface allows the user to monitor the bicycle’s status in real-time and switch between operating modes (Remote, Manual, and Autonomous). It also features integrated map-based navigation using OpenStreetMap.

For real-time balance control, a custom STM32-based PCB was developed to sense the roll angle and adjust the steering angle accordingly. The target heading (bearing) is calculated on the Jetson using a computer vision algorithm (OpenCV) and transmitted to the STM32 controller.

This system successfully demonstrated that the bicycle could autonomously navigate to a destination while maintaining self-balance, both with and without a rider.

Recognition: In recognition of its innovation, the system received multiple honours at POSTECH’s Spring 2018 and Spring 2019 Creative IT Design Competitions.

Additionally, the project won the Grand Prize (Minister’s Award) and the Excellence Award from the ICT Creative Consilience Programme at the Creative ICT Convergence Forums in 2018 and 2019.