A task-oriented, trajectory planner: Could it train stroke survivors to move normally on ADLs?

When real objects are used in therapy their presence affects movement kinematics and hence the trajectories themselves. We examine the effectiveness of a trajectory planning algorithm to model the natural curving wrist movements observed in real life functional tasks. The fifth order minimum jerk model that incorporates curvature is used as a starting point for the model. The model was further developed and modified using observed data from able-bodied individuals. The model will be implemented as a trajectory planning tool for the ADLER environment where it could be used to train stroke survivors during functional tasks. We assessed differences between predicted and actual movements for medium to high functional stroke survivors on a drink task for several variables such as peak velocity, total displacement, movement smoothness, etc. We found that the model was capable of predicting movements smoothness and total displacement but not peak velocities in the horizontal or vertical planes of movement. Our next steps would be to improve and implement the model in the ADLER environment for stroke rehabilitation.