Sub-Millisecond Reflexes for Continuous Physical Human-Robot Interaction
Authors: Jon Škerlj, Sami Haddadin
Organizations: Munich Institute of Robotics and Machine Intelligence (MIRMI), Technische Universität München (TUM), Germany. · Mohamed Bin Zayed University of Artificial Intelligence, UAE
Continuous physical human-robot interaction requires robots to rapidly adapt their control behaviour without interrupting task execution. However, existing ROS 2 control architectures rely on stop-to-switch mechanisms that introduce substantial delays and interrupt execution. We present an open-source, sub-millisecond reflex engine based on an enhanced ROS 2-compatible multimode controller that enables deterministic switching between control modalities and integrates reflex behaviours for mitigating contact hazards and kinematic limit violations. Experiments on a Franka Robotics manipulator demonstrate switching delays below 0.1 ms, more than four orders of magnitude faster than conventional ros2_control, while maintaining continuous operation during safety-critical events. In a multi-stage manipulation task, the proposed framework reduced execution time by 3.5% and peak trajectory jerk by more than threefold. By eliminating runtime middleware overhead, controller transitions are implemented as lock-free pointer operations, enabling reflex-level responsiveness within an open-source robotics framework.
Figures & tables
Fig. 1 : Conventional control frameworks interrupt tasks during control mode switching and hazardous events (left), whereas the proposed framework enables sub-millisecond reflex responses and uninterrupted physical human-robot interaction (right).
Fig. 2 : Visualization of the improved multimodal controller within the ros2_control framework. More detail about ros2_control components can be found in [ 20 ] .
Fig. 3 : State machine for reactive reflex behaviours.
TABLE I : Available state-of-the-art reflex behaviours [ 10 ] .
Fig. 4 : Box plots of controller switching delays using a logarithmic scale for a) baseline ros2_control framework, b) original multimode controller, and c) improved multimode controller for 100 switches per each configuration.
Fig. 5 : Stopping distances displayed per joint for a) joint position test and b) joint velocity test.
Fig. 6 : Example trajectories of Joint 2 for (a) joint position and (b) velocity limit tests. Recorded joint angles are shown in blue and velocities in black, corresponding to the respective colour-coded vertical axes.
Fig. 7 : Results of the contact detection tests displaying time to mitigate force after an unexpected collision.
Fig. 8 : Results of the contact detection tests displaying the magnitudes of end-effector force (blue) and velocity (black). Both trajectories correspond to the respective color-coded vertical axes.
Fig. 9 : Multi-stage task velocity comparison between state-of-the-art baseline ( ros2_control ) in red and our approach (enhanced multimode controller) in blue.
Metric
Proposed approach
ros2_control
Total Task Duration [s]
14.09±0.06
14.58±0.10
Peak Trajectory Jerk [m/s3]
155.4±40.6
499.0±96.9
TABLE II : Quantitative performance comparison across 20 trials of the multifaceted task ( N=10 per configuration).
Munich Institute of Robotics and Machine Intelligence (MIRMI), Technische Universität München (TUM), Germany · Technische Universität Nürnberg (UTN), Germany · Mohamed Bin Zayed University of Artificial Intelligence, UAE