Kinova Jaco wheelchair robot arm and hand for rehabilitation and assistance to the disabled

点击翻译按钮获取中文摘要

Robot arms and hands can potentially help disabled and elderly with daily manipulation tasks. Power wheelchairs are common today and provide mobility, and thus freedom and independence to people who would otherwise be home bound. Our research aims to give upper-body disabled individuals the ability to live a more independent life. Many users of power wheelchairs also have reduced arm function. Here robot arms and hands can help. However, having more joints and motors, these are far more complex to control than the wheelchair. We propose to research from issues in motion and force control to human-robot interfaces. The group of four co-PI span three departments, Computer Science, Electrical Engineering and Rehabilitation Medicine and are thus well qualified to take on this interdisciplinary research. To perform realistic experiments we need to purchase a wheelchair mounted robot arm and and hand. Kinova's Jaco robot is specifically designed to be light and slender so as to mount on the armrest of a wheelchair. Currently the Jaco arm and hand is controlled through the wheelchair joystick. However while a 2Degree-Of-Freedom (DOF) wheelchair joystick is sufficient for controlling the speed and direction of the two wheel motors in a wheelchair, it is cumbersome when used to control the 7 motors in the Jaco arm and hand. The user has to switch which spatial directions and rotations that the joystick controls numerous times in order to complete a manipulation task. Our goal is to use vision and other sensing to simplify the control interface to the robot. Computer vision can detect objects; once the user selects what objects they intend to grasp, the system could then calculate appropriate motion trajectories to these objects. In this approach, reachable objects would be shown on a screen and the user merely selects what object he or she wants to manipulate. The robot arm can then use the calculated trajectory to perform the detailed motion control to move towards the selected object. This greatly simplifies the user interaction from numerous joystick motions and mode switches to a single selection. Electromyography (EMG) provides the potential to control the robot also for individuals with other motor impairments who cannot use a physical interaction device. Currently the Jaco is position controlled. To enable nimbler interactions with the environment we intend to research force control based not on conventional end effector sensors, but on joint torque sensors and arm modeling. The wheelchair arm and hand system is portable and will be shared between the four PI between labs in Science, Medicine, and Engineering. We will also bring it to the Glenrose Rehabilitation hospital for patient tests.