Soft Robotics Workshop: Harvard & UTEC in Peru

This video introduces a polynomial curvature approximation method for modeling soft robots, where the robot's backbone curvature is represented as a polynomial function of position, enabling the transformation of the infinite-dimensional soft robot dynamics into a finite-dimensional system suitable for model-based control design; the approach demonstrates that using the constant curvature term as control output creates a minimum phase system, and proposes a PD-poly controller that incorporates high-order deformation information to achieve zero steady-state error in the presence of gravity and nonconstant curvature conditions.

Soft robotics is an emerging field that uses compliant materials (like silicone, rubber, and elastomers) instead of rigid metals to create robots capable of large deformations and safe human interaction. Unlike traditional robots with discrete joints, soft robots have infinite degrees of freedom, making their control more challenging due to non-linear behavior. Control strategies include pneumatic actuators (pneunets) with pressure-based control, cascaded control systems with inner pressure loops and outer position loops, and fuzzy logic controllers that handle the non-linearities of compliant materials. The control system typically uses sensors (like capacitive or resistive strain sensors) to measure deformation and feedback to controllers (such as PID or fuzzy controllers) to achieve desired positions or forces.

Soft robots encompass four main classes with distinct control characteristics: (1) Elastic joint robots use harmonic drives with inherent compliance, decoupling motor and link inertia for safer human interaction; (2) Variable stiffness actuation allows online modification of joint stiffness for performance-safety trade-offs; (3) Distributed flexibility in links creates non-collocation problems where torque at the base affects tip motion through complex dynamics; (4) Continuum soft robots have infinite degrees of freedom requiring special modeling approaches. The reduced model separates link equations (rigid dynamics with elastic coupling) from motor dynamics (linear with diagonal inertia). Distributed flexibility creates non-minimum phase behavior where conventional inverse dynamics becomes unstable, requiring specialized control frameworks that leverage rather than eliminate flexibility.

Soft robotics introduces the concept of infinite degrees of freedom because flexible materials can bend and twist in countless ways beyond discrete joint angles. Unlike rigid robots with fixed joint positions, soft robotic hands can conform to object shapes continuously. This flexibility provides unique advantages for grasping irregular objects but requires sophisticated control strategies to achieve precise manipulation despite the apparent lack of discrete control points.

Controlling soft robots requires a model-based approach that addresses their unique challenges: these continuum robots have infinite-dimensional configurations requiring discretization into finite degrees of freedom, exhibit underactuation where fewer actuators control more degrees of freedom, and possess inherent elastic and dissipative properties that can be leveraged for stabilization. The solution involves two key transformations: first discretizing the infinite-dimensional system into ordinary differential equations using curvature-based or piecewise constant curvature models, and second applying a change of coordinates to achieve collocated control where actuated coordinates directly respond to inputs. This enables precise shape regulation and trajectory tracking through PD controllers with explicit compensation for elastic and gravitational effects, outperforming model-free approaches in both steady-state accuracy and transient response.

Soft robotics has transformative applications in biomedicine. The Skiff Flop is a flexible endoscopic surgical robot enabling minimally invasive operations in complex conditions, developed through collaboration between Sant'Anna Institute of Pisa and the Italian Institute of Technology. Soft prosthetics, 3D printed and completely flexible, connect to the forearm using electrodes and can adapt force output for delicate or strong tasks. Hydrogel, a water-based biomaterial, offers exceptional biocompatibility for internal tissue contact, enabling microchirurgia applications. These advances demonstrate how soft robotics can address human needs in ways rigid robots cannot, from surgical precision to adaptive prosthetic functionality.

Soft Robotics offers significant advantages in medical applications due to mechanical compliance and biological tissue compatibility. Traditional hard robotics (metals) may trigger immune responses, be too bulky, or lack flexibility. Applications include physiological monitoring sensors that adhere to skin like stickers, surgical soft robotic systems for internal operations, and wearable devices for motor assistance benefiting stroke patients and workers in strenuous environments. Soft robotics enables adjustable casts that can be reshaped for different body parts and reused rather than discarded. The field intersects with biohybrid engineering, combining synthetic soft materials with actual biological tissue to enable self-repairing and adaptive implants like hip replacements that can repair themselves and last longer.

Soft robotics represents a paradigm shift in robot design, moving away from rigid mechanical systems toward flexible, adaptive structures. The elephant trunk-inspired manipulator from Oxford University uses shape-memory alloy wires that bend in any direction through heating, achieving the highest force-to-weight ratio among actuators. The 12-gram design enables delicate manipulation tasks. In biomedical applications, Oxford researchers developed technology for growing human cells on robotic scaffolds, combining skeletal models with nutrient media in biodegradable fiber membranes. The cells undergo daily mechanical loading to simulate physiological conditions. Japanese researchers created a robotic finger covered with living skin that can heal itself using collagen bandages, representing a significant step toward truly human-like robots despite challenges in public acceptance.

Soft robotics is revolutionizing healthcare through applications in prosthetics, surgery, and diagnostics. Prosthetic hands modeled after human anatomy offer safety, affordability, and scalability for pediatric patients. In minimally invasive surgery, soft robotic devices demonstrate superior dexterity and safety compared to rigid surgical robots, enabling natural navigation around organs and deeper tissue access. The STIFF-FLOP project validated these advantages through human cadaver studies. Additionally, soft robotic probes equipped with miniature tactile sensors enable real-time organ palpation during surgery, providing critical diagnostic information about tissue abnormalities. For theranostic applications, growing soft robots can navigate confined spaces like the colon, representing the next generation of endoscopic technology that combines therapeutic and diagnostic capabilities in a single flexible platform.

Harvard University, in collaboration with Trinity College Dublin, has created an open-source Soft Robotics Toolkit (softroboticstoolkit.com) that provides free intellectual resources—including downloadable plans, tutorials, sample code, and a collaborative database—for enthusiasts and researchers to build soft robots at home using common materials like silicone and Arduino boards, enabling applications in rescue operations, prosthetics, and adaptive clothing.

Open source hardware democratizes access to technology by freely sharing design files and manufacturing instructions. This approach enables global collaboration where individuals worldwide can source components locally and assemble devices. Crowdfunding models extend reach beyond local communities, while user feedback drives continuous improvement. The PiKon project exemplifies this: starting as a single demonstration, it evolved into a global movement with users modifying designs, adding features like motorization, and deploying telescopes in underserved communities.

Soft robotics is an emerging field that combines classical robot design principles with soft, flexible materials to create low-cost devices capable of achieving complex motions through inflation-based actuation; the Soft Robotics Toolkit provides open-source resources including CAD files for mold design, tutorials for fabrication, modeling guidance, and an open-source control board to enable students, designers, and researchers to learn and develop novel soft robotic devices.

Open hardware is an emerging paradigm that makes hardware designs freely available for everyone to study, modify, and redistribute, similar to how open source software transformed information sharing; this approach, combined with affordable 3D printing technology, enables global collaboration to solve real-world problems—from creating low-cost prosthetic hands for children to monitoring water quality and environmental conditions—demonstrating that collective innovation can empower individuals and communities to address complex challenges more effectively than traditional proprietary approaches.