Publications

The Duke Humanoid: Design and control for energy-efficient bipedal locomotion using passive dynamics Authors: B. Xia¹, B. Li¹, J. Lee¹, M. Scutari¹, B. Chen¹ ¹Duke University Publication: IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2025) (Accepted) Presentation: Poster DOI: 10.48550/arXiv.2409.19795 (arXiv preprint) Abstract We present the Duke Humanoid, an open-source 10-degrees-of-freedom humanoid, as an extensible platform for locomotion research. The design mimics human physiology, with symmetrical body alignment in the frontal plane to maintain static balance with straight knees. We develop a reinforcement learning policy that can be deployed zero-shot on the hardware for velocity-tracking walking tasks. Additionally, to enhance energy efficiency in locomotion, we propose an end-to-end reinforcement learning algorithm that encourages the robot to leverage passive dynamics. Our experimental results show that our passive policy reduces the cost of transport by up to 50% in simulation and 31% in real-world tests. Our website is https://github.com/generalroboticslab/DukeHumanoidv1. ...

Text2Robot: Evolutionary robot design from text descriptions Authors: R.P. Ringel¹, Z.S. Charlick¹, J. Liu¹, B. Xia¹, B. Chen¹ ¹Duke University Publication: 2025 IEEE International Conference on Robotics and Automation (ICRA), pp. 5789–5797 (2025) Presentation: Poster DOI: 10.1109/ICRA55743.2025.11128168 Abstract Robot design has traditionally been costly and labor-intensive. Despite advancements in automated processes, it remains challenging to navigate a vast design space while producing physically manufacturable robots. We introduce Text2Robot, a framework that converts user text specifications and performance preferences into physical quadrupedal robots. Within minutes, Text2Robot can use text-to-3D models to provide strong initializations of diverse morphologies. Within a day, our geometric processing algorithms and body-control co-optimization produce a walking robot by explicitly considering real-world electronics and manufacturability. Text2Robot enables rapid prototyping and opens new opportunities for robot design with generative models. Our website is at http://generalroboticslab.com/Text2Robot/. ...

A legged soft robot platform for dynamic locomotion Authors: B. Xia¹, J. Fu¹, H. Zhu¹, Z. Song¹, Y. Jiang¹, H. Lipson¹ ¹Columbia University Publication: 2021 IEEE International Conference on Robotics and Automation (ICRA), pp. 11812–11819 (2021) Presentation: Poster DOI: 10.1109/ICRA48506.2021.9561018 Abstract This paper presents an open-source untethered quadrupedal soft robot platform for dynamic locomotion (e.g., high-speed running and backflipping). The robot is mostly soft (80 vol.%) while driven by four geared servo motors. The robot’s soft body and soft legs were 3D printed with gyroid infill using a flexible material, enabling it to conform to the environment and passively stabilize during locomotion in multi-terrain environments. In addition, we simulated the robot in a real-time soft body simulation. With tuned gaits in simulation, the real robot can locomote at a speed of 0.9 m/s (2.5 body length/second), substantially faster than most untethered legged soft robots published to date. We hope this platform, along with its verified simulator, can catalyze agile soft robots’ development. ...

Improving the actuation speed and multi-cyclic actuation characteristics of silicone/ethanol soft actuators Authors: B. Xia¹, A. Miriyev¹, C. Trujillo¹, N. Chen¹, M. Cartolano¹, S. Vartak¹, H. Lipson¹ ¹Columbia University Publication: Actuators, 9(3), 62 (2020) Cover Article DOI: 10.3390/act9030062 Abstract The actuation of silicone/ethanol soft composite material-actuators is based on the phase change of ethanol upon heating, followed by the expansion of the whole composite, exhibiting high actuation stress and strain. However, the low thermal conductivity of silicone rubber hinders uniform heating throughout the material, creating overheated damaged areas in the silicone matrix and accelerating ethanol evaporation. This limits the actuation speed and the total number of operation cycles of these thermally-driven soft actuators. In this paper, we showed that adding 8 wt.% of diamond nanoparticle-based thermally conductive filler increases the thermal conductivity (from 0.190 W/mK to 0.212 W/mK), actuation speed and amount of operation cycles of silicone/ethanol actuators, while not affecting the mechanical properties. We performed multi-cyclic actuation tests and showed that the faster and longer operation of 8 wt.% filler material-actuators allows collecting enough reliable data for computational methods to model further actuation behavior. We successfully implemented a long short-term memory (LSTM) neural network model to predict the actuation force exerted in a uniform multi-cyclic actuation experiment. This work paves the way for a broader implementation of soft thermally-driven actuators in various robotic applications. ...

Additive manufacturing of silicone composites for soft actuation Authors: A. Miriyev¹, B. Xia¹, J.C. Joseph¹, H. Lipson¹ ¹Columbia University Publication: 3D Printing and Additive Manufacturing, 6(6), 309–318 (2019) Cover Article DOI: 10.1089/3dp.2019.0116 Abstract Elastomer composites have proven to be promising functional materials for soft actuators. Direct manufacturing of these materials is a practical prerequisite for Soft Robotics applications, where form and function are intricately entangled. In this article we show a multimaterial printer and associated processes for in situ fabrication of silicones and silicone-based elastomer composites for soft actuators. We discuss the fabrication process for both the silicone/ethanol composite material and encapsulating silicone skin using an inline passive mixing system, followed by characterization of the rheological and mechanical properties of the printed materials for various print modalities. Rheological study revealed the conditions, allowing continuous 3D printing of both small and big items out of silicone rubber and silicone/ethanol composite. Anisotropic mechanical properties allow for the design of functional characteristics of soft actuators by choosing print design modalities. We demonstrate a single-print-job additive manufacturing of functional multimaterial systems for soft actuation, and suggest that the developed processes will allow us to design soft robots with a broad range of actuation characteristics. ...

Development of a virtual floor maze test – effects of distal visual cues and correlations with executive function in healthy adults Authors: D. Martelli¹, A. Prado¹, B. Xia¹, J. Verghese², S.K. Agrawal¹,³ ¹Department of Mechanical Engineering, Columbia University ²Department of Neurology, Albert Einstein College of Medicine ³Department of Rehabilitation and Regenerative Medicine, Columbia University Publication: IEEE Transactions on Neural Systems and Rehabilitation Engineering, 27(10), 2229–2236 (2019) DOI: 10.1109/TNSRE.2019.2938103 Abstract This study presents the development of a virtual reality version of the floor maze test (VR-FMT), a navigational assessment for evaluating spatial navigation through an unfamiliar two-dimensional floor maze. Fifty-five healthy adults completed the floor maze test in three conditions (real environment, virtual environment without walls, and virtual environment with walls) and two neuropsychological tests. The study examined how distal visual cues affect maze navigation performance and correlations with executive function. Results showed that the VR-FMT provides an ecologically valid environment for assessing spatial navigation abilities and correlates with executive function measures. This work offers potential applications for early detection and remediation of cognitive and navigational deficits in clinical populations. ...

Conductive fabric heaters for heat-activated soft actuators Authors: M. Cartolano¹, B. Xia¹, A. Miriyev¹, H. Lipson¹ ¹Columbia University Publication: Actuators, 8(1), 9 (2019) DOI: 10.3390/act8010009 Abstract We examine electrically conductive fabrics as conductive heaters for heat-activated soft actuators. We have explored various fabric designs optimized for material properties, heat distribution and actuation/de-actuation characteristics of the soft actuators. We implemented this approach in the silicone/ethanol composite actuators, in which ethanol undergoes a thermally-induced phase change, leading to high actuation stress and strain. Various types of conductive fabrics were tested, and we developed a stretchable kirigami-based fabric design. We demonstrate a fabric heater that is capable of cyclic heating of the actuator to the required 80 °C. The fabric with the special kirigami design can withstand temperatures of up to 195 °C, can consume up to 30 W of power, and allows the actuator to reach >30% linear strain. This technology may be used in various systems involving thermally-induced actuation. ...

Gait adaptations during overground walking and multidirectional oscillations of the visual field in a virtual reality headset Authors: D. Martelli¹*, B. Xia¹*, A. Prado¹, S.K. Agrawal¹,² *Equal contribution ¹Department of Mechanical Engineering, Columbia University ²Department of Rehabilitation and Regenerative Medicine, Columbia University Publication: Gait & Posture, 67, 251–256 (2019) DOI: 10.1016/j.gaitpost.2018.10.029 Abstract Virtual reality (VR) has been used to study locomotor adaptability during balance-demanding tasks, but little is known about gait adaptations during overground walking. This study investigated how twelve healthy young adults modified and adapted their gait during overground walking in a virtual environment (VE) with continuous multidirectional perturbations of the visual field while wearing a VR headset. Participants walked on an instrumented walkway in four conditions: real environment, VE, and VE with antero-posterior and medio-lateral visual perturbations. Results showed that participants took shorter strides with higher variability in the VE and during perturbations, with medio-lateral perturbations causing greater effects. Over time, participants adapted by increasing stride length and reducing variability. This paper provided first evidence for visuomotor adaptations during overground walking with a VR headset, representing an initial investigation for developing new VR methods for early detection and remediation of gait deficits in ecological conditions. ...