Physical and cognitive disabilities make it difficult or impossible to perform simple personal or job-related tasks. The primary objective of this research and development effort is to assist persons with physical disabilities to perform activities of daily living (ADL) using a smart 9-degrees-of-freedom (DOF) modular wheelchair-mounted robotic arm system (WMRA).

The combination of the wheelchair's 2-DoF mobility control and the robotic arm's 7-DoF manipulation control in a single control mechanism allows people with disabilities to do many activities of daily living (ADL) tasks that are otherwise hard or impossible to accomplish. Different optimization methods for redundancy resolution are explored and modified to fit the new system with combined mobility and manipulation control and to accomplish singularity and obstacle avoidance as well as other optimization criteria to be implemented on the new system. The resulting control algorithm of the system is tested in simulation using C++ and Matlab codes to resolve any issues that might occur during the testing on the physical system. Implementation of the combined control is done on the newly designed robotic arm mounted on a modified power wheelchair and with a custom designed gripper.

The user interface is designed to be modular to accommodate any user preference, including a haptic device with force sensing capability, a spaceball, a joystick, a keypad, a touch screen, head/foot switches, sip and puff devices, and the BCI 2000 that reads the electromagnetic pulses coming out of certain areas of the brain and converting them to control signals after conditioning.

Different sensors (such as a camera, proximity sensors, a laser range finder, a force/torque sensor) can be mounted on the WMRA system for feedback and intelligent control. The user should be able to control the WMRA system autonomously or using teleoperation. Wireless Bluetooth technology is used for remote teleoperation in case the user is not on the wheelchair. Pre-set activities of daily living tasks are programmed for easy and semi-autonomous execution.