The impressive capabilities of humans to robustly perform manipulation relies on compliant interactions, enabled through the structure and materials spatially distributed in our hands. We propose by mimicking this distributed compliance in an anthropomorphic robotic hand, the open-loop manipulation robustness increases and observe the emergence of human-like behaviours. To achieve this, we introduce the ADAPT Hand equipped with tunable compliance throughout the skin, fingers, and the wrist. Through extensive automated pick-and-place tests, we show the grasping robustness closely mirrors an estimated geometric theoretical limit, while `stress-testing' the robot hand to perform 800+ grasps. Finally, 24 items with largely varying geometries are grasped in a constrained environment with a success rate of 93%. We demonstrate the hand-object self-organization behavior underlines this extreme robustness, where the hand automatically exhibits different grasp types depending on object geometries. Furthermore, the robot grasp type mimics a natural human grasp with a direct similarity of 68%.
Large Language Models can lead researchers in the design of robots.
Aquatic creatures exhibit remarkable adaptations of their body to efficiently interact with the surrounding fluid. The tight coupling between their morphology, motion, and the environment are highly complex but serves as a valuable example when creating biomimetic structures in soft robotic swimmers. We focus on the use of asymmetry in structures to aid thrust generation and maneuverability. Designs of structures with asymmetric profiles are explored so that we can use morphology to `shape' the thrust generation. We propose combining simple simulation with automatic data-driven methods to explore their interactions with the fluid. The asymmetric structure with its co-optimized morphology and controller is able to produce 2.5 times the useful thrust compared to a baseline symmetric structure. Furthermore these asymmetric feather-like arms are validated on a robotic system capable of forward swimming motion while the same robot fitted with a plain feather is not able to move forward.
The success of soft robots in displaying emergent behaviors is tightly linked to the compliant interaction with the environment. However, to exploit such phenomena, proprioceptive sensing methods which do not hinder their softness are needed. In this work we propose a new sensing approach for soft underwater slender structures based on embedded pressure sensors and use a learning-based pipeline to link the sensor readings to the shape of the soft structure. Using two different modeling techniques, we compare the pose reconstruction accuracy and identify the optimal approach. Using the proprioceptive sensing capabilities we show how this information can be used to assess the swimming performance over a number of metrics, namely swimming thrust, tip deflection, and the traveling wave index. We conclude by demonstrating the robustness of the embedded sensor on a free swimming soft robotic squid swimming at a maximum velocity of 9.5 cm/s, with the absolute tip deflection being predicted within an error less than 9% without the aid of external sensors.
Soft robot are celebrated for their propensity to enable compliant and complex robot-environment interactions. Soft robotic manipulators, or slender continuum structure robots have the potential to exploit these interactions to enable new exploration and manipulation capabilities and safe human-robot interactions. However, the interactions, or perturbations by external forces cause the soft structure to deform in an infinite degree of freedom (DOF) space. To control such system, reduced order models are needed; typically models consider piecewise sections of constant curvature although external forces often deform the structure out of the constant curvature hypothesis. In this work we perform an analysis of the trade-off between computational treatability and modelling accuracy. We then propose a new kinematic model, the Piecewise Affine Curvature (PAC) which we validate theoretically and experimentally showing that this higher-order model better captures the configuration of a soft continuum body robot when perturbed by the external forces. In comparison to the current state of the art Piecewise Constant Curvature (PCC) model we demonstrate up to 30\% reduction in error for the end position of a soft continuum body robot.
In this review we introduce the framework of reality-assisted evolution to summarize a growing trend towards combining model-based and model-free approaches to improve the design of physically embodied soft robots. In silico, data-driven models build, adapt and improve representations of the target system using real-world experimental data. By simulating huge numbers of virtual robots using these data-driven models, optimization algorithms can illuminate multiple design candidates for transference to the real world. In reality, large-scale physical experimentation facilitates the fabrication, testing and analysis of multiple candidate designs. Automated assembly and reconfigurable modular systems enable significantly higher numbers of real-world design evaluations than previously possible. Large volumes of ground-truth data gathered via physical experimentation can be returned to the virtual environment to improve data-driven models and guide optimization. Grounding the design process in physical experimentation ensures the complexity of virtual robot designs does not outpace the model limitations or available fabrication technologies. We outline key developments in the design of physically embodied soft robots under the framework of reality-assisted evolution.