Self-driving wheelchairs have been developed, but they have not reached the market yet as many of them depend on some propriety hardware of the wheelchair, which changes as manufacturers come up with new designs. Additionally, the power wheelchairs available in the market use a standard joystick module– an input device that is unsuitable for our client due to involuntary muscle spasms. To overcome these challenges, this thesis presents the development of a modular control software architecture for navigating a power wheelchair, with minimal adaptation needed for the underlying power wheelchair hardware. Given the user's requirement to use an eye-controlled tablet, the software prototype was designed to integrate this assistive device for wheelchair control. This project is funded by Ability Tech (Lab), which develops smart assistive devices for people with disabilities.
The prototype named Power-Wheelchair-Add-On Control-System-Unit (PWAO CSU) was implemented and tested in this work. The PWAO CSU software architecture is structured in three layers: the user interface layer, the sequence control layer built on the ROS 2 framework, and the loop control layer. This architecture provides interfaces for the different control layers, establishing communication between the power wheelchair and the user's assistive device. The prototype also features a PID controller for precise turning motion and a simple obstacle avoidance system for safety. The functionality and real-time performance of the system were tested. Additionally, a user test was conducted to gain insights into user experience. While the user was able to use the eye-controlled tablet to manoeuvre the power wheelchair, a performance bottleneck was identified in the communication between the assistive device and the ROS 2 sub-system. Additionally, based on other findings, several recommendations were suggested to improve its functionality and performance.