Abstract:
Powered wheelchairs significantly improve mobility of individuals with physical limitations,
but their high cost and challenges associated with traditional joystick controls
often limit their accessibility and ease of use. This thesis aims to explore alternative
modern input methods for controlling a standard powered wheelchair aiming to
redesing the system for shared control implementation. A ground-up approach is
taken, covering wheelchair electronics modification, localization, path planning, and
comparison of different input methods.
The research primarily focuses on the development and evaluation of a experimental
prototype based on an Otronica Pulse 310 electric wheelchair with a ZED2
camera and ROS installed laptop. Localization performance was optimized using Simultaneous
Localization and Mapping (SLAM) by combining the ZED2 camera with
encoders. Caster wheel-aware path planning was achieved using Model Predictive
Control (MPC) and potentiometers mounted directly on the axis. Various user input
methods, such as touchscreen control and shared control, were investigated, which
proved to be more comfortable and user-friendly than traditional joystick controls.
Despite the limited time available for implementation, the study provides a solid
foundation for future research on wheelchair input methods, path-planning, and
control systems. It also offers valuable insights for enhancing the lives of powered
wheelchair users through improved control interfaces, shared control, and navigation
technologies.