Ruckig

Open-source library that generates jerk-limited, time-optimal trajectories in real time, every control cycle.

Ruckig is an open-source C++ library (with a Python interface) by Lars Berscheid, with its paper published at RSS 2021. Given the current state (position, velocity, acceleration), a target state, and each joint's limits on velocity, acceleration, and jerk (the rate of change of acceleration), it computes a time-optimal trajectory in microseconds — fast enough to re-plan on every single control cycle. That means when the target suddenly changes, the robot can smoothly transition from whatever motion state it's currently in, which is the “online” part of its name. MoveIt 2 uses it for trajectory smoothing and limiting in real-time servoing. The core is open source, with a commercial Ruckig Pro edition adding features such as intermediate waypoints.

ExampleDuring visual servoing, the target position changes every frame; the controller calls Ruckig every 1 ms to re-plan from the current velocity and acceleration to the new target, so the joints don't jerk when the target jumps.

Also called

Ruckig Pro

Related

Jerk · S-Curve Velocity Profile · Time Parameterization · Time-Optimal Path Parameterization · TOPP-RA · MoveIt Motion Planning Framework

Sources

pantor/ruckig (GitHub)
Jerk-limited Real-time Trajectory Generation with Arbitrary Target States (arXiv)