Soft robots (SRs) represent one of the most significant recent evolutions in robotics. Designed to embody safe and natural behaviors, they rely on compliant physical structures purposefully designed to embody desirable and sometimes variable impedance characteristics. This article discusses the problem of controlling SRs. We start by observing that most of the standard methods of robotic control-e.g., high-gain robust control, feedback linearization, backstepping, and active impedance control-effectively fight against or even completely cancel the physical dynamics of the system, replacing them with a desired model. This defeats the purpose of introducing physical compliance. After all, what is the point of building soft actuators if we then make them stiff by control?

Despite the classic nature of the problem, trajectory tracking for soft robots, i.e., robots with compliant elements deliberately introduced in their design, still presents several challenges. One of these is to design controllers which can obtain sufficiently high performance while preserving the physical characteristics intrinsic to soft robots. Indeed, classic control schemes using high-gain feedback actions fundamentally alter the natural compliance of soft robots effectively stiffening them, thus de facto defeating their main design purpose. As an alternative approach, we consider here using a low-gain feedback, while exploiting feedforward components. In order to cope with the complexity and uncertainty of the dynamics, we adopt a decentralized, iteratively learned feedforward action, combined with a locally optimal feedback control. The relative authority of the feedback and feedforward control actions adapts with the degree of uncertainty of the learned component. The effectiveness of the method is experimentally verified on several robotic structures and working conditions, including unexpected interactions with the environment, where preservation of softness is critical for safety and robustness.

Legged robotic technologies have moved out of the lab to operate in real environments, characterized by a wide variety of unpredictable irregularities and disturbances, all this in close proximity with humans. Demonstrating the ability of current robots to move robustly and reliably in these conditions is becoming essential to prove their safe operation. Here, we report an in-depth literature review aimed at verifying the existence of common or agreed protocols and metrics to test the performance of legged system in realistic environments. We primarily focused on three types of robotic technologies, i.e., hexapods, quadrupeds and bipeds. We also included a comprehensive overview on human locomotion studies, being it often considered the gold standard for performance, and one of the most important sources of bioinspiration for legged machines. We discovered that very few papers have rigorously studied robotic locomotion under irregular terrain conditions. On the contrary, numerous studies have addressed this problem on human gait, being nonetheless of highly heterogeneous nature in terms of experimental design. This lack of agreed methodology makes it challenging for the community to properly assess, compare and predict the performance of existing legged systems in real environments. On the one hand, this work provides a library of methods, metrics and experimental protocols, with a critical analysis on the limitations of the current approaches and future promising directions. On the other hand, it demonstrates the existence of an important lack of benchmarks in the literature, and the possibility of bridging different disciplines, e.g., the human and robotic, towards the definition of standardized procedures that will boost not only the scientific development of better bioinspired solutions, but also their market uptake.

This work proposes a method to grasp unknown objects with robotic hands based on demonstrations by a skilled human operator. Not only are humans efficacious at grasping with their own hands but are also capable of grasping objects using robotic hands. Therefore, we consider how the grasping skills of a human trained in robotic hand use can be transferred to a robot equipped with the same hand. We propose that a skilled human user manually operates the robotic hand to grasp a number of elementary objects, consisting of different boxes. A Decision Tree Regressor is trained on the data acquired from the human operator to generate hand poses able to grasp a general box. This is extended to grasp objects of unknown and general shape leveraging upon the state-of-theart Minimum Volume Bounding Box decomposition algorithm that approximates with a number of boxes the point cloud of the object. We report on extensive tests of the proposed approach on a Panda manipulator equipped with a Pisa/IIT SoftHand, achieving a success rate of 86.7% over 105 grasps of 21 different objects.

In this letter, we aim to improve the robustness of dynamic quadrupedal locomotion through two aspects: 1) fast model predictive foothold planning, and 2) applying LQR to projected inverse dynamic control for robust motion tracking. In our proposed planning and control framework, foothold plans are updated at 400 Hz considering the current robot state and an LQR controller generates optimal feedback gains for motion tracking. The LQR optimal gain matrix with non-zero off-diagonal elements leverages the coupling of dynamics to compensate for system underactuation. Meanwhile, the projected inverse dynamic control complements the LQR to satisfy inequality constraints. In addition to these contributions, we show robustness of our control framework to unmodeled adaptive feet. Experiments on the quadruped ANYmal demonstrate the effectiveness of the proposed method for robust dynamic locomotion given external disturbances and environmental uncertainties.

In this paper, we first discuss the challenges related to habitat monitoring and review possible robotic solutions. Then, we propose a framework to perform terrestrial habitat monitoring exploiting the mobility of legged robotic systems. The idea is to provide the robot with the Natural Intelligence introduced as the combination of the environment in which it moves, the intelligence embedded in the design of its body, and the algorithms composing its mind. This approach aims to solve the challenges of deploying robots in real natural environments, such as irregular and rough terrains, long-lasting operations, and unexpected collisions, with the final objective of assisting humans in assessing the habitat conservation status. Finally, we present examples of robotic monitoring of habitats in four different environments: forests, grasslands, dunes, and screes.