I am currently Taking this Class!

Come back at the end of the semester for more!

Project Overview:

We plan to create a real-time planner for a 7 DoF robotic manipulator that minimizes the chances of encountering singularities (i.e. maximizes dexterity) while performing physical rehabilitation based on observed and predicted human motion trajectories. The predicted trajectories will be calculated from inverse kinematics of the idealized human with a probabilistic model used to represent the wide range of human movement patterns. The planner must account for locations of singularities and obstacles in the environment (such as the human subject or the links of the robot itself) and plan accordingly to avoid them. We will use dexterity measures [1,2] and the safety margin from the human as objective functions to control the robot while maximizing dexterity and keep the human safe within the workspace. (the lower level controller for the robot is either impedance or admittance control and hence inherently safe due to the torque sensors at the joints)