A Unified Fractal Impedance Control Framework
TL;DR Packages Fractal Impedance Control into one open-source library and GUI that run the same controllers in MuJoCo and on real arms, and ports the H-PMP architecture, which nests task-space and joint-space admittance control, to a Kinova Gen3. H-PMP tracks a figure-eight with 7.8 mm RMSE against 12.6-43.2 mm for earlier FIC controllers, and overshoots least (23.4 mm) when a wall it presses on suddenly drops away.
What they do
Fractal Impedance Control (FIC) is a passive, asymptotically stable interaction controller whose variants have so far lived in separate implementations. This paper packages them into one C++ library with ROS2, MuJoCo and hardware drivers, plus a GUI called FIC Studio. It also gives the first full robot implementation of the Harmonic-Passive Motion Paradigm (H-PMP), an architecture built to model human motor control. H-PMP trades off admittance and impedance behaviour without a state machine.
How it works
- Library and Studio. Robots are described by configuration files, and the same library and tuning profile run in MuJoCo and on hardware (tested on a Kinova Gen3 and a Franka FR3). Effort profiles are retuned live by dragging handles, without a restart.
- HapFIC. An admittance controller whose offset is bounded by the FIC saturations, which gives the trade-off between admittance and impedance.
- H-PMP. A task-space admittance or HapFIC controller turns the desired pose and external wrench into desired joint torques for a joint-space HapFIC. Inverse kinematics supplies the joint posture, and gravity is compensated separately.
- Setup. 1 kHz torque control on a Kinova Gen3, with the external wrench estimated from joint torques.
Results
- Figure-eight tracking (10 trials each): H-PMP RMSE 7.8 mm, against 12.6 mm (Cartesian impedance with joint-space HapFIC), 21.1 mm (Task-Space HapFIC) and 43.2 mm (Cartesian impedance with planner).
- Wall press with sudden release: H-PMP holds 12.6 N and travels 23.4 mm past the wall face, against 36.3 mm for Task-Space HapFIC (4.6 N) and 68.7 mm for Cartesian impedance with joint-space HapFIC (16.9 N).
- Kinesthetic teaching then replay: H-PMP replays with 40.7 mm RMSE against 55.7 mm for the baseline, and 42.4 mm with its joint layer on an STM32 microcontroller.
Limitations
Only FIC-based controllers are compared, and the wall and teaching experiments use one or two trials per controller. H-PMP’s path is less smooth than the purely Cartesian controllers’ and it has higher orientation error than the Cartesian impedance ones (3.33° against 1.41-1.94°). On its first press it rebounded from the wall once. The authors expect online tuning to improve performance.
Figures
Abstract
Physical human--robot interaction requires controllers capable of maintaining intrinsically bounded interactions with the environment. The Fractal Impedance Controller (FIC) has been proven to guarantee these safety characteristics by design. This paper introduces the robotic implementation and experimental validation of its most advanced architecture: the Harmonic-Passive Motion Paradigm (H-PMP), an evolution of the Passive Motion Paradigm designed to mimic human-like interaction. Furthermore, we present a unified framework for FIC deployment, supported by an open-source software suite featuring ROS2, MuJoCo, and physical hardware integrations. Through three experiments conducted on a Kinova Gen3 arm (trajectory tracking, a wall press with sudden loss of contact, and learning-by-demonstration) we compare the H-PMP against earlier FIC architectures. The results demonstrate that the H-PMP achieves a seamless trade-off between admittance and impedance behaviours. This makes it ideal for deployment in unstructured environments, enabling the robot to transition between interaction modes with no computational overhead.
Summaries are written automatically from each paper's text. Figures and videos come from the paper and its project page; all credit goes to the authors.