Automatic Adaptation of an Humanoid Robot Gait to Different Floor-Robot Friction Coefficients



    Project name: Automatic Adaptation of an Humanoid Robot Gait to Different Floor-Robot Friction Coefficients
    PI: João Paulo Morais Ferreira
    Reference: PTDC/EEI-AUT/5141/2014
    Fundind entity: FCT
    Partners: Portugal (ISR, IPC, UA), Japan (KUT) and Serbia (MPI)
    Duration of the action: 2016-07-01 - 2019-06-30
    Description: A human gait trajectory acquisition will be performed using an upgraded acquisition system already developed using 2 digital video cameras, one to the sagittal and the other to the lateral plane. It will be also developed an instrumented pair of shoes for reading the vertical and horizontal forces on a human walking shoe, allowing the calculation of shoe-floor friction coefficient and CoP trajectories. Another innovation proposed by the project team is to include the reinforcement learning to automatically adapt the robot walking when it detects the change of the friction condition between the robot and the ground. Finally, to complement the complex dynamic model of the humanoid robot, a haptic system (implemented using a force feedback device) will be developed to allow real time feedback of the robot's CoP to an operator to correct its stability, changing the ankle and torso angles manually. Data from this system will be used to train the SVR (Support Vector Regression) stability control system of the humanoid robot. It is expected that the instrumented shoes developed can also be used for medical applications, allowing gait pathologies identification and quantification of their severity. It can be an important new medical diagnosis tool, overcoming the cost limitations of the present gait diagnosis tools. This system would allow an objective understanding of the clinical evolution of patients, enabling an effective functional rehabilitation of a patient's gait. This system will be tested in the Physical and Rehabilitation Medicine Unit, Coimbra Hospital and University Centre (CHUC).

Associated People





You are here: Home People
Website Security Test