Research Of Real-time Obstacle Avoidance And Shared Control Of A Smart Walking-aid Robot

Since the 21 st century, the world's population gradually presents the trend of aging. Meanwhile, frequent accidents, diseases and disasters make the disabled people increase day by day. These people face degeneration and loss of body function, so they need rehabilitation training to maintain or recover their body function. But the existing rehabilitation is difficult to meet their needs, so it is important and urgent to research and develop smart rehabilitation appliances and achieve productization around the world.The research of rehabilitation robot has some imperfections at present, such as insufficient consideration on users' safety, poor human-robot interaction, single function and so on. For this situation, this thesis develops a smart walking-aid robot to help the elderly and the disabled who still have a certain ability to walk with their lower limb rehabilitation training. The thesis makes an innovation in the human-robot interaction interface of the walking-aid robot: by the rational distribution of a set of one dimension force sensors, the user's intention can be accurately identified with lower cost, a laser range finder(LRF) is also used to let the walking-aid robot get the ability of real-time obstacle avoidance. Obstacle avoidance algorithm is designed to improve the safety of the system, artificial potential field is applied to calculate the repulsion from the obstacles around the walking-aid robot, a standard path that has attraction on the walking-aid robot is also set. When there are obstacles in the walking-aid robot's emergency area, the emergency obstacle avoidance algorithm can help to avoid collision accident. Admittance control is also applied to convert the forces applied to the walking-aid robot to its kinematic velocity smoothly. Combined with the human-robot interaction interface, both the user's intentioned destination and his intentioned walking pattern can be identified. Finally distance between the walking-aid robot and obstacles, the deviation from the standard path, the ratio of its current velocity and the max velocity and the average velocity of the users are considered in shared control algorithm to let the walking-aid robot follow the user's intention as much as possible under the premise of user's safety, making the user's operation more stable and more comfortable.Three experiments are designed to prove the effectiveness of the design and algorithms above. Intention identification experiment is firstly done, by letting different volunteers drive the walking-aid robot to the given destination and switch their walking pattern many times, the accuracy of human-robot interaction interface and admittance control algorithm on user's intention identification is proved. Then a volunteer is asked to simulate a patient with lower limb disabled and visually impaired, he is also asked to do the experiment of obstacle avoidance and the experiment of shared control in the same indoor environment. From the change of the related parameters during the experiment and the comparison of the path, the effectiveness of the obstacle avoidance algorithm and the improvement of the control effect by using shared control algorithm can be proved.In summary, the walking-aid robot developed by this thesis has gained a certain researching achievement in human-robot interaction interface design, the identification of the user's intention, obstacle avoidance control and shared control. This thesis has made a contribution to the productization of walking-aid robots.