We want to be able to follow both joint space and cartesian space trajectories. Usually, these will be produced using cubics.
It is likely possible to achieve this by sampling the produced trajectories and sending these to the Dynamixel while in the time-based profile.
We need to experiment with how to achieve these. We can try putting the Dynamixel in the time-based profile and playing with the velocity and acceleration values while sending it sample points from the trajectory. We can then plot the planned trajectory and the actual Dynamixel trajectory in the same graph. The goal is to get these close to each other.
To start, it makes sense to only use one of the Dynamixels (e.g. the wrist joint or the x-y axis one).
We want to be able to follow both joint space and cartesian space trajectories. Usually, these will be produced using cubics. It is likely possible to achieve this by sampling the produced trajectories and sending these to the Dynamixel while in the time-based profile. We need to experiment with how to achieve these. We can try putting the Dynamixel in the time-based profile and playing with the velocity and acceleration values while sending it sample points from the trajectory. We can then plot the planned trajectory and the actual Dynamixel trajectory in the same graph. The goal is to get these close to each other. To start, it makes sense to only use one of the Dynamixels (e.g. the wrist joint or the x-y axis one).
Also, see here.