Sebin Jung

Real-time whole-body controllers for legged robots typically plan through a fixed nominal model and degrade when the deployed dynamics change. Existing adaptive methods either require model structure that contact dynamics do not provide or need offline training for each anticipated condition. We present Look-back and Look-ahead Adaptive Model Predictive Path Integral control (LLA-MPPI). The method instead converts whole-body adaptation into selection over a bank of GPU-batched contact simulators with different physical or structural parameters. Windowed prediction errors select the simulator that best explains recent motion. A whole-body MPPI planner then optimizes controls through the selected model. The framework requires no offline training and its identified parameters remain physical quantities. Across four simulated tasks it achieves 97.5% success while the strongest baseline reaches 74% and an oracle given the true model reaches 98.5%. Hardware experiments on a Unitree Go2 show the robot walking under a payload added mid-run, walking after one leg is disabled, and pushing a box to its goal while the box mass is increased on the fly.

Robots operating in unstructured environments may fall or overturn, and recovering an upright posture is critical for sustained operation. Inspired by the feline righting reflex, in which coordinated body and tail motions contribute to aerial reorientation, this work investigates the role of spine and tail actuation in aerial self-righting. We hypothesized that adding an actuated tail improves self-righting performance despite its added mass and mechanical complexity. To evaluate this hypothesis, we present Meow-Mentum, a 1.86 kg self-righting robot that combines a two-degree-of-freedom (2-DoF) actuated spine with a one-degree-of-freedom (1-DoF) actuated tail, using onboard sensing and closed-loop control. Reinforcement learning-based controllers were trained in a physics simulator and deployed zero-shot on the physical robot. Meow-Mentum achieved feet-down orientation within 1 m of free fall from a range of initial orientations. With active spine motion in both morphologies, we compared the matched tail and no-tail configurations across pitch and roll release conditions. Adding tail actuation improved self-righting performance overall and across roll releases, while its additional benefit across pitch releases was less clear. These results indicate that the contribution of tail actuation to self-righting depends on the initial release orientation.

Controlling robots with strongly nonlinear, high-dimensional dynamics remains challenging, as direct nonlinear optimization with safety constraints is often intractable in real time. The Koopman operator offers a way to represent nonlinear systems linearly in a lifted space, enabling the use of efficient linear control. We propose a data-driven framework that learns a Koopman embedding and operator from data, and integrates the resulting linear model with the Safe Set Algorithm (SSA). This allows the tracking and safety constraints to be solved in a single quadratic program (QP), ensuring feasibility and optimality without a separate safety filter. We validate the method on a Kinova Gen3 manipulator and a Go2 quadruped, showing accurate tracking and obstacle avoidance.

@misc{jung2026wholebodysafecontrolrobotic,
                title = {Whole-Body Safe Control of Robotic Systems with Koopman Neural Dynamics},
                author = {Sebin Jung and Abulikemu Abuduweili and Jiaxing Li and Changliu Liu},
                year = {2026},
                eprint = {2603.03740},
                archivePrefix = {arXiv},
                primaryClass = {cs.RO},
                url = {https://arxiv.org/abs/2603.03740}
              }

S. Jung, A. Abuduweili, J. Li and C. Liu, "Whole-Body Safe Control of Robotic Systems with Koopman Neural Dynamics," arXiv:2603.03740, 2026.

This paper introduces the Safe Protective and Assistive Robot Kit (SPARK), a comprehensive benchmark designed to ensure safety in humanoid autonomy and teleoperation. Humanoid robots pose significant safety risks due to their physical capabilities of interacting with complex environments. The physical structures of humanoid robots further add complexity to the design of general safety solutions. To facilitate the safe deployment of complex robot systems, SPARK can be used as a toolbox that comes with state-of-the-art safe control algorithms in a modular and composable robot control framework. Users can easily configure safety criteria and sensitivity levels to optimize the balance between safety and performance. To accelerate humanoid safety research and development, SPARK provides a simulation benchmark that compares safety approaches in a variety of environments, tasks, and robot models. Furthermore, SPARK allows quick deployment of synthesized safe controllers on real robots. For hardware deployment, SPARK supports Apple Vision Pro (AVP) or a Motion Capture System as external sensors, while also offering interfaces for seamless integration with alternative hardware setups. This paper demonstrates SPARK’s capability with both simulation experiments and case studies with a Unitree G1 humanoid robot. Leveraging these advantages of SPARK, users and researchers can significantly improve the safety of their humanoid systems as well as accelerate relevant research.

@inproceedings{sun2025spark,
                title = {SPARK: Safe Protective and Assistive Robot Kit},
                author = {Sun, Yifan and Chen, Rui and Yun, Kai S and Fang, Yikuan and Jung, Sebin and Li, Feihan and Li, Bowei and Zhao, Weiye and Liu, Changliu},
                booktitle = {IFAC Symposium on Robotics},
                year = {2025},
                url = {https://arxiv.org/pdf/2502.03132}
              }

Y. Sun, R. Chen, K. Yun, Y. Fang, S. Jung, F. Li, B. Li, W. Zhao and C. Liu, "SPARK: Safe Protective and Assistive Robot Kit" in IFAC Symposium on Robotics, 2025.