Beyond Rolling: A Tensegrity Robot Adapts to Unconventional Movement

Author: Denis Avetisyan

Researchers have developed a novel tensegrity robot capable of dynamic locomotion through internal mass shifting and surprisingly resilient behavior.

Mobile tensegrity robots employing curved-link designs demonstrate significantly greater locomotion speed-when normalized to body lengths per second-than those built with straight-link architectures, as evidenced by comparisons across five published experimental platforms [6, 7, 8, 9, 10].

This work details the design, simulation, and experimental validation of TeXploR2, a tensegrity robot utilizing non-intuitive control strategies for adaptable and impact-resistant locomotion.

Conventional mobile robots often trade speed for stability when navigating unstructured terrain, while highly effective rolling designs can lack the robustness needed for complex environments. This limitation motivates the research presented in ‘When Rolling Gets Weird: A Curved-Link Tensegrity Robot for Non-Intuitive Behavior’, which introduces TeXploR2, a tensegrity robot employing curved links and internal mass shifting to achieve unexpectedly efficient and controllable locomotion. Through simulation and experimental validation-including impact testing demonstrating inherent shock absorption-we show that dynamic rolling can be achieved via non-intuitive control strategies. Could this approach unlock new possibilities for adaptable robots operating in challenging, real-world scenarios?

Beyond Wheels: Embracing Adaptability in Robotic Locomotion

Conventional wheeled robots, while effective on smooth, predictable surfaces, struggle significantly when confronted with the irregularities of real-world terrains. Their rigid structure transmits impact forces directly to the chassis, potentially damaging sensitive components and hindering operational efficiency. This lack of compliance-the ability to deform and adapt to uneven ground-limits their maneuverability over obstacles like rocks, stairs, or even thick vegetation. Furthermore, the fixed axle configuration often leads to wheel slippage and loss of traction, reducing their ability to navigate complex landscapes or maintain stability on inclines. These limitations underscore the need for alternative locomotion methods that prioritize adaptability and robust performance in challenging environments, moving beyond the constraints of traditional wheeled designs.

Current robotic locomotion often relies on rigid wheeled systems, which struggle with uneven or unpredictable surfaces. Researchers are increasingly focused on designs that embrace compliance – the ability to deform and adapt to external forces – and actively manage how weight is distributed during movement. This shift prioritizes dynamic stability over static rigidity, allowing robots to absorb impacts, navigate obstacles, and maintain balance even when confronted with challenging terrain. By mimicking the natural flexibility of biological systems, these new approaches promise more robust and versatile robots capable of operating effectively in real-world environments where perfect surfaces are rarely encountered. The goal isn’t simply to overcome obstacles, but to redistribute weight in real-time, essentially ‘feeling’ the terrain and adjusting to it, paving the way for truly adaptable robotic movement.

TeXploR2 successfully demonstrated dynamic rolling locomotion on a treadmill and concluded with a stable impact test on a rubber mat.

The Core Innovation: Internal Mass Shifting

Internal mass shifting utilizes the principle of moving weighted components along curved pathways – specifically arcs – to directly induce rolling motion in a robotic system. This contrasts with conventional wheeled or legged locomotion by eliminating the need for external contact with the ground. The system achieves propulsion through the cyclical redistribution of mass, creating an internal imbalance that drives forward movement. By strategically shifting weights along these arcs, the robot effectively ‘rolls’ without relying on friction or external forces, representing a fundamentally different approach to terrestrial locomotion and potentially offering advantages in terms of efficiency and maneuverability.

The robot’s locomotion relies on a tensegrity structure built from semi-circular, rigid links. This design prioritizes both stability and speed through the distribution of forces across the structure. Tensegrity, characterized by isolated compression members within a network of continuous tension, allows the robot to maintain structural integrity even with dynamic weight shifts. The semi-circular links specifically contribute to efficient force transfer during rolling, minimizing energy loss and maximizing the robot’s ability to achieve speeds up to 1.88 body lengths per second. This structural approach also provides a degree of compliance, enabling the robot to navigate uneven terrain while maintaining balance.

Locomotion is achieved through the controlled displacement of internal masses, necessitating the implementation of NEMA23 Stepper Motors to ensure precision and reliability in their operation. These motors facilitate dynamic adjustments to the mass distribution, currently enabling a rolling speed of up to 1.88 body lengths per second. The NEMA23 specification provides sufficient torque and step resolution for accurate positioning of the weights along their defined arcs, directly correlating to the achieved velocity and responsiveness of the system. This motor selection is critical for maintaining stability and efficiently translating mass movement into propulsive force.

The TeXploR2 prototype, significantly larger and heavier than its predecessor, utilizes internal mass shifting and a network of 12 elastic cables to achieve rolling locomotion via CoM manipulation.

From Simulation to Reality: Validating the Model

A MATLAB simulation was developed to model the quasistatic rolling sequence of the robot. This simulation facilitated a detailed analysis of mass movement throughout the rolling cycle, quantifying the distribution of mass at discrete points in time. Force distribution was also modeled, calculating the forces exerted on the contact points between the robot and the rolling surface. The simulation environment allowed for parametric studies to assess the impact of various design and operational parameters on the robot’s rolling behavior, informing optimization strategies for both efficiency and stability. Data generated from the simulation included positional data of mass elements, velocity profiles, and reaction forces, all of which served as benchmarks for experimental validation.

Simulation results indicated that complete mass transfer to the arc endpoint during the quasistatic rolling sequence is not the optimal strategy for robot locomotion. Analysis revealed that retaining a portion of the mass closer to the robot’s center of gravity enhances both efficiency and stability. This non-intuitive behavior minimizes energy expenditure by reducing the required force for subsequent movements and improves stability by lowering the center of mass and decreasing the risk of tipping during the rolling process. This finding challenges conventional assumptions regarding complete mass transfer in robotic locomotion systems.

Experimental validation of the quasistatic rolling sequence simulation was performed using Vicon motion capture technology. This system utilizes infrared cameras to track the three-dimensional position of markers affixed to the robot, allowing for precise measurement of its movement during the rolling process. The Vicon system achieved a positional accuracy of 0.3 millimeters, enabling a high degree of confidence in the correlation between the simulated and physical robot behavior and providing data for refining the model’s parameters.

A quasistatic locomotion simulation demonstrates a rolling sequence-traversing states 1 to 4 to 2 to 3-as indicated by colored lines representing movement and dots/crosses denoting contact points and the center of mass, respectively.

A Functional Prototype: Demonstrating Resilient Locomotion

The culmination of this research is a tetherless prototype that achieves locomotion through the innovative application of internal mass shifting. This system eschews traditional propulsion methods, instead relying on the precisely controlled movement of a substantial mass-up to 1150 grams-within the robot’s chassis to generate forward motion. By strategically shifting this weight along a dedicated track, the prototype effectively alters its center of gravity, creating an imbalance that drives movement. This approach not only demonstrates the feasibility of internal mass shifting as a viable locomotion strategy, but also highlights its potential for creating robots capable of navigating complex terrains and absorbing significant impacts – confirmed through rigorous testing – without relying on wheels, legs, or external forces.

The robot’s robust stability is fundamentally achieved through a carefully engineered mechanical system. A precision T-Track serves as the guiding infrastructure for the stepper motors, ensuring accurate and controlled internal mass movement along a linear path. Complementing this is the implementation of V-Groove Bearings, which effectively minimize any potential carriage rocking during locomotion. These bearings maintain consistent contact between the moving mass and the robot’s frame, preventing unwanted oscillations and contributing to a remarkably smooth and controlled gait, even across uneven terrain. This synergistic combination of guiding track and stabilizing bearings allows for precise weight shifting, a critical element in achieving dynamic and resilient movement.

Rigorous impact testing revealed the robot’s surprising ability to absorb and dissipate kinetic energy, demonstrating a notable resilience crucial for operation in unpredictable terrains. Researchers subjected the prototype – utilizing an internal mass shift of 1150 grams and a curved arc mass totaling 1300 grams – to a series of controlled impacts, observing minimal disruption to functionality even after significant force was applied. This inherent shock absorption isn’t achieved through traditional dampers or flexible materials, but rather through the dynamic redistribution of mass itself, effectively lowering the center of gravity upon impact and mitigating the transmission of force to sensitive components. The results suggest a design capable of withstanding considerable external stresses, paving the way for deployment in environments where robustness is paramount.

The motor carriage design utilizes v-groove and ball bearing attachments to a T-track, stabilizing the carriage and preventing rocking even at high velocities.

Future Directions: Towards Intelligent and Adaptive Robots

The incorporation of an Inertial Measurement Unit (IMU) represents a significant advancement in robotic navigation and control. This sensor suite, providing continuous data on the robot’s orientation and acceleration, allows for real-time adjustments to maintain stability and optimize movement, even across unpredictable terrain. By precisely tracking changes in angular velocity and linear acceleration, the robot can anticipate and react to disturbances, enhancing its ability to navigate complex environments and avoid collisions. This proactive response, facilitated by the IMU’s data, moves beyond reactive obstacle avoidance, enabling smoother, more efficient, and ultimately, more intelligent robotic locomotion and manipulation capabilities.

The implementation of feedback control algorithms promises a paradigm shift in robotic locomotion, allowing for dynamic mass redistribution in response to varying terrains and external disturbances. These algorithms continuously monitor sensor data – including ground reaction forces and robot orientation – and then strategically shift internal mass to maintain balance and optimize stability. For example, as the robot encounters sloped terrain, the control system can autonomously shift mass towards the downward slope, lowering the center of gravity and preventing tipping. Similarly, when subjected to an external force – such as a push or an impact – the system can rapidly redistribute mass to counteract the force and maintain equilibrium. This adaptive mass shifting not only enhances robustness and prevents falls but also improves energy efficiency by minimizing the need for excessive joint actuation, paving the way for robots capable of navigating complex and unpredictable environments with remarkable agility and resilience.

The implementation of Onyx material represents a significant step towards bolstering the long-term operational capacity of the robot. Traditional robotic components often suffer from fatigue and fracture under repeated stress, limiting their lifespan and requiring frequent maintenance. Onyx, however, exhibits markedly improved durability and flexural strength – meaning it can withstand considerably greater forces and bending without permanent deformation or failure. This enhanced material resilience translates directly into a more reliable robotic platform, capable of enduring the rigors of complex terrains and prolonged use. By minimizing the risk of structural compromise, Onyx not only extends the robot’s functional lifespan but also reduces the need for costly repairs and replacements, ultimately contributing to a more sustainable and efficient robotic system.

Using TeXploR, a dynamic rolling sequence was successfully reconstructed on a stationary treadmill, demonstrating winding roll patterns (represented by red-to-magenta and blue-to-cyan arcs) and culminating in a successful impact test on a rubber mat.

The presented research embodies a pursuit of elegant simplicity in robotic locomotion. TeXploR2, with its dynamic rolling and internal mass shifting, challenges conventional understandings of stability and control. This resonates with the sentiment expressed by John McCarthy: “The best way to predict the future is to invent it.” The robot doesn’t merely adapt to pre-defined terrains; it creates its own possibilities through non-intuitive movements. The core concept of resilient locomotion through compliant mechanisms isn’t about overcoming complexity, but about transcending it – a testament to the power of inventive design over brute force solutions. It’s a clear illustration that true progress often lies in reimagining fundamental principles.

Beyond Rolling

The demonstration of dynamic rolling locomotion in a tensegrity robot-TeXploR2-reveals less a solution, and more a repositioning of the problem. Resilience through impact is noted, but resilience to complexity remains unaddressed. Current geometric modeling, while functional, presumes a static ideal. A truly adaptable system necessitates models that acknowledge, even embrace, inherent imperfection.

Future iterations should not focus on achieving faster rolling, but on relinquishing control. The capacity for internal mass shifting suggests potential for self-stabilization, a system that responds to external forces not through programmed reaction, but through intrinsic re-configuration. This demands a shift in metrics. Efficiency is a vanity; robustness-measured by the system’s capacity to absorb and redistribute energy-is paramount.

The question is not how to build a robot that rolls predictably, but how to build one that accepts its own unpredictability. Clarity is the minimum viable kindness. Further work should explore the limits of this acceptance, acknowledging that the most elegant solutions are often the simplest – and the most forgiving of error.

Original article: https://arxiv.org/pdf/2603.16503.pdf

Contact the author: https://www.linkedin.com/in/avetisyan/