Tag Archives: planning-control-distributed

TRUSSES: Temporarily, Robots Unite to Surmount Sandy Entrapments, then Separate

This project is a collaboration with research groups at Penn (Mark Yim, Daniel Koditschek, Douglas Jerolmack) and at USC (Feifei Qian). The aim for this project is to develop methods for teams of robots to jointly overcome environmental hazards on the Moon by attaching to each other to form larger and more stable, maneuverable structures. The robots will use their interactions with the ground to form a map of safe and risky terrain, attach to each other as support when the ground traversal risk is high, move in a coordinated fashion once joined, and, once the maneuver has been successfully completed, separate to continue their original individual missions.

The TRUSSES project will develop algorithms that provide robots with two main capabilities: 1) estimate robot-to-regolith interactions in order to plan safe maneuvers, and 2) plan truss formations and coordinated motions for robots to push and pull each other to safe locations. The system will be evaluated to verify risk estimation, risk avoidance, risk mitigation, and recovery from failure.

Our group primarily develops planning and localization methods for coordinated, risk-aware maneuvers in heterogeneous teams of ground robots operating with a partially known map. This map is built proprioceptively by an exploring quadruped and encodes spatial parameters for a robot-ground interaction model, enabling force prediction. It can also be transformed into a risk map that reflects uncertainty (unexplored regions) and poor terrain (where robots may get stuck). Heterogeneity is advantageous, as each robot interacts differently with the terrain, offering unique capabilities in team operations. This diversity requires high-level planning to assign roles aligned with each robot’s strengths, producing team configurations that support the rescue mission while avoiding high-risk zones. Once a target configuration is defined, robots reposition themselves using a reactive navigation planner and form truss connections. A truss-planner then computes joint motions so the connected network can move collectively to safer areas, where the robots can disconnect and resume individual tasks like exploration or transport.

Related Publications

Scout-rover cooperation: Online terrain strength mapping and traversal risk estimation for planetary-analog explorations

Liu, Shipeng; Caporale, J. Diego; Fulcher, Ethan; Hu, Wilson; Cavallo, Natalie; Zhang, Yifeng; Liao, Xingue; Sung, Cynthia; Qian, Feifei

Scout-rover cooperation: Online terrain strength mapping and traversal risk estimation for planetary-analog explorations (Conference)

Lunar and Planetary Science Conference (LPSC), 2025.

(BibTeX)

TRUSSES: Temporarily, Robots Unite to Surmount Sandy Entrapments, then Separate

Jerolmack, Douglas; Koditschek, Daniel; Qian, Feifei; Yim, Mark; Sung, Cynthia

TRUSSES: Temporarily, Robots Unite to Surmount Sandy Entrapments, then Separate (Conference)

Lunar Surface Innovation Consortium Spring Meeting, 2024.

(BibTeX)

Current Personnel

  • Rachel Holladay (MEAM Postdoc)
  • Shivangi Misra (ESE PhD)
  • William Hoganson (ROBO Staff)
  • Eric Wang (CIS Undergrad)
  • Lori Brown (CIS Undergrad)
  • Neha Peddinti (CIS Undergrad)
  • Wilson Hu (CIS Undergrad)
  • Benjamin Aziel (MEAM Master's)

Past Personnel

  • Mason Mitchell (CIS Staff)
  • Jun Kwon (MEAM, CS Undergrad)
  • Paul Young (MEAM Master's)

Acknowledgments

The work is supported by Lunar Surface Technology Research grant #80NSSC24K0127 from NASA’s Space Technology Research Grants Program. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of this organization.

Tunable Stiffness in Soft Robots

The emergence of smart actuators with tunable material properties has empowered roboticists to create systems capable of transitioning between rigid and soft operational modes. This adaptability allows robots to operate with high precision and enhanced payload capacity in their rigid state, while offering greater safety and flexibility when soft especially in the presence of unpredictable disturbances. We design these adaptive robotic systems and develop corresponding control algorithms to showcase the versatility and potential of this new class of reconfigurable robots.

Fig 1: Collage of projects in the research thrust: Tunable Stiffness in Soft Robots

Coiled Spring Actuator

We have developed several novel tunable-stiffness actuators as part of this research thrust, one of which is the coiled-spring actuator. It is inspired by the mechanics of nested elastic rings, wherein the effective bulk stiffness can be modulated by varying the number of elastic coiled layers. This mechanism enables near-linear stiffness tuning, achieved through electro-mechanical control. The system allows for precise, programmable stiffness adjustments, and our prototype has demonstrated a tunability range of up to 20-fold. We have also constructed a non-dimensional mechanics model for the coiled spring actuator which extends to all such mechanisms of different dimensions and materials. The local stiffness changes from these actuators induce corresponding deformations in a compliant, segmented manipulator constructed as a tower of multiple such modules.

Pneumatic Dual-Bellows Actuator

Another novel tunable stiffness actuator developed in our lab is able to achieve a stiffness gain of  1.43 times (1332 N/m to 1913 N/m) without needing an external air source or valve. The design consists of an air chamber bellows and spring bellows connected to each other in an air-tight manner. Stiffness modulation in the spring bellows is achieved by altering the volume of the air chamber bellows. Due to large achievable stiffnesses, this actuator is suitable for integration in soft robots that are needed to demonstrate dynamic and adaptable behavior.

Hierarchical Algorithms to Optimize Soft Manipulator Mechanics

Stiffness control algorithms are needed for soft manipulators to be able to take effective advantage of embedded novel stiffness actuators. To address this, we have developed a hierarchical policy for stiffness control for a class of soft segmented manipulators. The stiffness changes induce desired deformations in each segment, thereby influencing the manipulator’s end-effector position. The algorithm can be run as an online controller to influence the manipulator’s stable states or offline as a design algorithm to optimize stiffness distributions.

Coupled Learning in Elastic Networks

We are also collaborating with Lauren Altman and Doug Durian in the Physics Department, School of Arts and Sciences, University of Pennsylvania to build an elastic network for coupled learning. This technique tunes the properties of individual elastic elements in the network to achieve specific outcomes, like applying the right force or strain to an output edge. These mechanical networks could be scaled and automated to tackle more complex tasks, opening the door to a new kind of smart metamaterial.

Related Publications

Online Optimization of Soft Manipulator Mechanics via Hierarchical Control

Misra, Shivangi; Sung, Cynthia

Online Optimization of Soft Manipulator Mechanics via Hierarchical Control (Conference)

7th IEEE-RAS International Conference on Soft Robotics (RoboSoft), 2024.

(Abstract | BibTeX | Links: )

Design and Characterization of a Pneumatic Tunable-Stiffness Bellows Actuator

Chen, Rongqian; Kwon, Jun; Chen, Wei-Hsi; Sung, Cynthia

Design and Characterization of a Pneumatic Tunable-Stiffness Bellows Actuator (Conference)

IEEE-RAS International Conference on Soft Robotics (RoboSoft), 2024.

(Abstract | BibTeX | Links: )

Design and Control of a Tunable-Stiffness Coiled-Spring Actuator

Misra, Shivangi; Mitchell, Mason; Chen, Rongqian; Sung, Cynthia

Design and Control of a Tunable-Stiffness Coiled-Spring Actuator (Conference)

IEEE International Conference on Robotics and Automation (ICRA), 2023.

(Abstract | BibTeX | Links: )

Forward kinematics and control of a segmented tunable-stiffness 3-D continuum manipulator

Misra, Shivangi; Sung, Cynthia

Forward kinematics and control of a segmented tunable-stiffness 3-D continuum manipulator (Conference)

IEEE International Conference on Robotics and Automation (ICRA), 2022.

(Abstract | BibTeX | Links: )

Current Personnel

  • Wei-Hsi Chen (ESE Postdoc)
  • Shivangi Misra (ESE PhD)

Past Personnel

  • Mason Mitchell (CIS Staff)
  • Jun Kwon (MEAM, CS Undergrad)
  • Allen Zhou (MEAM Master's)
  • Kefan Wu (ESE Master's)
  • Rongqian Qi (ESE Master's)
  • Ruizhe Wang (ESE Master's)

Acknowledgments

Support for these projects has been provided by the National Science Foundation (NSF) Grant No. 1845339, the Johnson & Johnson WiSTEM2D Scholars Program, and the Army Research Office (ARO) under MURI award #W911NF1810327. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of funding source.

SALP: Salp-inspired Approach to Low-Energy Propulsion

Jet propulsion is a locomotion mode commonly found in biological swimmers, including cephalopods and tunicates such as squids, cuttlefish, and salps. We are developing a soft salp-inspired robotic system to study mechanisms that produce greater locomotion agility and energetic efficiency.

Salps are barrel-shaped marine invertebrates that swim via jet propulsion. They move forward by rapidly changing the volumes of their body cavity, drawing water into their muscular mantle cavity through the front aperture, and then expelling it under high pressure through the rear funnel. We aim to leverage the unique biomechanics of salps to inform the development of energy-efficient, maneuverable underwater robots capable of environmental sensing in complex marine environments.

Salps can swim either as solitary jet-propelled individuals or while physically connected in a multi-jet colony, commonly known as a “salp chain”. Inspired by salps, we develop the SALP (Salp-inspired Approach to Low-energy Propulsion) robot, a soft underwater robot that swims via jet propulsion similarly to a biological salp.

Version 1: Origami Swimmer

Traditional soft robots use flexible materials for shape change but require complex fabrication. To simplify this, we used an origami-inspired design, which allows the robot to fold from flat sheets into 3D shapes, making it easier to store, transport, and assemble in just a few hours. The robot mimics squid locomotion using the origami magic ball pattern, which transforms between an ellipsoid and a sphere. This enables it to expand and contract, pulling in and expelling water to create a jet for propulsion. A tendon mechanism in its spine controls its length. Ongoing effects focus on evaluating different configurations’ effect (e.g. adding a front nozzle) on the performance of the robot and understanding the underlying dynamics of it.

Mechanism demo

Robot swimming

Version 2: Multi-Robot SALP Robot Platform for Long-Term Distributed Sensing

We have optimized the jetting robot design to enable higher thrust and lower drag. The SALP robot is now modular in the sense that the robots can be manually attached to each other in different physical arrangements to study the effect of multi-robot interactions on locomotion performance. Physically connected SALP chains coordinate their jets to achieve various propulsion modes. We are interested in investigating how physical arrangement and jet coordination between two robots affect the swimming performance and energy efficiency of a two-SALP robotic system. We aim to gain insights into the potential hydrodynamic benefits of multi-jet propulsion by exploring how different coordination strategies influence the surrounding flow environment.

Leveraging Fluid-Structure Interactions for Efficient Control in Geophysical Flows

Micro-vehicles are cost-effective platforms for robotics and automation, excelling in maneuverability and adaptability in diverse environments. However, their lightweight and limited computational capacity pose control challenges. Using our underwater platform, we aim to understand fluid-structure interactions to enhance design and control, resulting in more efficient micro-vehicles with extended lifespans. This effort is a collaborative project focusing on fluid dynamics, control theory, and reconfiguration planning. The project aims to leverage environmental forces for power efficiency, investigating morphological adaptations and passive transport properties. It seeks to synthesize motion control strategies considering inertial effects and fluid-structure interactions while exploring efficiency trade-offs. Ultimately, it aims to enhance micro-autonomous vehicles’ capabilities for long-term operations and future large-scale deploy.

Related Publications

Effect of Jet Coordination on Underwater Propulsion with the Multi-Robot SALP System

Yang, Zhiyuan; Zhang, Yipeng; Herbert, Matthew; Hsieh, M. Ani; Sung, Cynthia

Effect of Jet Coordination on Underwater Propulsion with the Multi-Robot SALP System (Conference)

8th IEEE-RAS International Conference on Soft Robotics (RoboSoft 2025), Forthcoming.

(Abstract | BibTeX | Links: )

Drag coefficient characterization of the origami magic ball

Chen, Guanyu; Chen, Dongsheng; Weakly, Jessica; Sung, Cynthia

Drag coefficient characterization of the origami magic ball (Proceedings Article)

In: ASME International Design Engineering Technical Conferences and Computers and Information in Engineering Conference (IDETC/CIE), pp. DETC2023-117182, 2023.

(Abstract | BibTeX | Links: )

Origami-inspired robot that swims via jet propulsion

Yang, Zhiyuan; Chen, Dongsheng; Levine, David J.; Sung, Cynthia

Origami-inspired robot that swims via jet propulsion (Journal Article)

In: IEEE Robotics and Automation Letters, vol. 6, no. 4, pp. 7145-7152, 2021.

(Abstract | BibTeX | Links: )

Current Personnel

  • Dongsheng Chen (MEAM PhD)
  • Zhiyuan (Annie) Yang (MEAM PhD)
  • Ryan Stanford (MEAM Undergrad)
  • Benedict Onyekwe (ROBO Master's)
  • Jingshuo Li (MEAM Master's)
  • Neel Mulay (MEAM Master's)

Past Personnel

  • Adithya Selvakumar (ESE Undergrad)
  • Guanyu Chen (MEAM Master's)
  • Yipeng Zhang (MEAM Master's)
  • Yunyi Chu (MEAM Master's)
  • Zhiyuan (Annie) Yang (MEAM, MCIT Master's)

Acknowledgments

The project “Leveraging Fluid-Structure Interactions for Efficient Control in Geophysical Flows” is in collaboration with Ani Hsieh’s lab from University of Pennsylvania, Eric Forgoston’s lab from Montclair State University, and Philip Yecko’s lab from The Cooper Union.

These projects have been supported by the National Science Foundation (NSF) Grant No. 2121887 and the Office of Naval Research (ONR) award #N00014-23-1-2068. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of funding source.