Due to the CoVID-19 pandemic, I was unable to proceed with the necessary experiments to prove that the solution provided by the trajectory planning algorithm, therefore I need to resort to the next best thing: simulations.
The initial results of the trajectory planning have proved to be able to produce the force necessary to achieve the goals of BETER REHAB, but further fine tunning of the controller is needed for an even smoother experience. To be able to fine tune the controller, I developed a simulated rehabilitation environment, described visually in the video bellow. Here are the elements included:
- A predefined ‘desired’ trajectory, which is what the patient would like to do. This is unknown to the robot.
- A simulated ‘intention of motion’. This tries to simulate the output of the intention of motion algorithm. It resembles the ‘desired’ trajectory, but it contains noise, the same way that the intention of motion algorithm outputs something that only resembles the real desired trajectory.
- A way to calculate the direction of the interaction force between the robot and the patient. This is considered to be along the vector defined by the difference between the current ‘desired’ position of the patient’s arm and the current ‘actual’ position of the robot.
We are now able to perform such a simulation using the gazebo model provided by the IIWA Stack. You can see a first demonstration on the video below
