This article tackles the problem of controlling articulated soft robots (ASRs), i.e., robots with either fixed or variable elasticity lumped at the joints. Classic control schemes rely on ...high-authority feedback actions, which have the drawback of altering the desired robot softness. The problem of accurate control of ASRs, without altering their inherent stiffness, is particularly challenging because of their complex and hard-to-model nonlinear dynamics. Leveraging a learned anticipatory action, iterative learning control (ILC) strategies do not suffer from these issues. Recently, ILC was adopted to perform position control of ASRs. However, the limitation of position-based ILC in controlling variable stiffness robots is that whenever the robot stiffness profile is changed, a different input action has to be learned. Our first contribution is to identify a wide class of ASRs, whose motion and stiffness adjusting dynamics can be proved to be decoupled. This class is described by two properties that we define: strong elastic coupling, relative to motors and links of the system and their connections; and homogeneity, relative to the characteristics of the motors. Furthermore, we design a torque-based ILC scheme that, starting from a rough estimation of the system parameters, refines the torque needed for the joint positions tracking. The resulting control scheme requires minimum knowledge of the system. Experiments on variable stiffness robots prove that the method effectively generalizes the iterative procedure with respect to the desired stiffness profile and allows good tracking performance. Finally, potential restrictions of the method, e.g., caused by friction phenomena, are discussed.
Since their introduction in the early years of this century, variable stiffness actuators (VSA) witnessed a sustained growth of interest in the research community, as shown by the growing number of ...publications. While many consider VSA very interesting for applications, one of the factors hindering their further diffusion is the relatively new conceptual structure of this technology. When choosing a VSA for their application, educated practitioners, who are used to choosing robot actuators based on standardized procedures and uniformly presented data, would be confronted with an inhomogeneous and rather disorganized mass of information coming mostly from scientific publications. In this paper, the authors consider how the design procedures and data presentation of a generic VSA could be organized so as to minimize the engineer’s effort in choosing the actuator type and size that would best fit the application needs. The reader is led through the list of the most important parameters that will determine the ultimate performance of their VSA robot, and influence both the mechanical design and the controller shape. This set of parameters extends the description of a traditional electric actuator with quantities describing the capability of the VSA to change its output stiffness. As an instrument for the end-user, the VSA datasheet is intended to be a compact, self-contained description of an actuator that summarizes all of the salient characteristics that the user must be aware of when choosing a device for their application. At the end some examples of compiled VSA datasheets are reported, as well as a few examples of actuator selection procedures.
•We used a quadrupedal robot with a Mobile Laser Scanning to structurally monitor EU forest habitats.•The robot's locomotion performance, efficiency, and accuracy were tested in EU Annex I Habitat ...9110.•Two scanning trajectories were used to map tree locations and extract tree DBH from point clouds.•Circular trajectory showed better tree detection rates, battery consumption and RMSE.•The RMLS survey significantly improved survey efficiency compared to traditional methods.
EU States are mandated by the 92/43/EEC Habitats Directive to generate recurring reports on the conservation status and functionality of habitats at the national level. This assessment is based on their floristic and, especially for forest habitats, structural characterization. Currently, habitat field monitoring efforts are carried out only by trained human operators. The H2020 Project “Natural Intelligence for Robotic Monitoring of Habitats – NI” aims to develop quadrupedal robots able to move autonomously in the unstructured environment of forest habitats. In this work, we tested the locomotion performance, efficiency and the accuracy of a robot performing structural habitat monitoring, comparing it with traditional field survey methods inside selected stands of Luzulo-Fagetum beech forests (9110 Annex I Habitat). We used a quadrupedal robot equipped with a Mobile Laser Scanning system (MLS), implementing for the first time a structural monitoring of EU forest habitats with a Robotically-mounted Mobile Laser Scanning (RMLS) platform. Two different scanning trajectories were used to automatically map individual tree locations and extract tree Diameter at Breast Height (DBH) from point clouds. Results were compared with field human measurements in terms of accuracy and efficiency of the survey. The robot was able to successfully execute both scanning trajectories, for which we obtained a tree detection rate of 100 %. Circular scanning trajectory performed better in terms of battery consumption, Root Mean Square Error (RMSE) of the extracted DBH (2.43 cm or 10.73 %) and prediction power (R2adj = 0.72, p < 0.001). The RMLS platform improved survey efficiency with 19.31 m2/min or 1.77 trees/min in comparison with 3.45 m2/min or 0.32 trees/min of traditional survey. Finally, a processing script was developed to allow the repeatability of RMLS surveys in similar habitat monitoring missions. In the future, a human-robotic monitoring framework might represent an accurate support for those repetitive and time-consuming activities in habitat monitoring, offering a valuable benefit for biodiversity conservation.
Despite the remarkable growth of the global market for robotics, robotic monitoring of habitats is still an understudied topic. This is true, among others, for the species-rich EU Annex I habitat ..."6210 - Semi-natural grasslands and scrubland facies on calcareous substrates". This habitat is typically surveyed by human operators. In this work, we present a dataset concerning relevés performed through the quadrupedal robot ANYmal C. The dataset contains information from three plots, which include the robot state, videos, and images acquired to assess the habitat conservation status. Additionally, a collection of videos and pictures about two typical and one early warning species of habitat 6210 is also presented. This database is publicly available in the provided Zenodo repository and will aid researchers in several fields. Robot state information can be used by engineers to validate their algorithms, while data gathered by the robot can be used to design new methodologies and new metrics to assess the habitat conservation status or train/test classifiers (e.g. neural networks) for plant classification.
Effective monitoring of habitats is crucial for their preservation. As the impact of anthropic activities on natural habitats increases, accurate and up-to-date information on the state of ecosystems ...has become imperative. This paper presents a new dataset collected from the forests located in the Tuscan Apennines (Italy) using the ANYmal robot. The dataset provides information regarding the structure and composition of the EU priority habitat 9210*. The dataset, which is publicly available through a Zenodo repository, includes photos, videos, and point clouds of the environment. This dataset is a valuable resource for the scientific community working in the field of forest ecology and conservation and has the potential to inform future research and conservation efforts on habitat 9210*. The collaboration between robotic engineers and plant scientists provides a unique perspective on the forest ecosystem and underscores the potential for interdisciplinary work in this field. This dataset constitutes an important contribution to the ongoing effort to monitor and conserve habitats globally, particularly in light of the challenges posed by global changes.
This data descriptor presents a novel dataset collected using the quadrupedal robot ANYmal C in the Mediterranean coastal dune environment of the European Union (EU) habitats 2110 and 2120 in ...Sardinia, Italy. The dataset mainly consists of photos, videos, and point clouds of the coastal dunes, providing valuable information on the structure and composition of this habitat. The data was collected by a team of robotic engineers and plant scientists as result of a joint effort towards robotic habitat monitoring. The dataset is publicly available through Zenodo and can be used by researchers working in both the fields of robotics and habitat ecology and conservation. The availability of this dataset has the potential to inform future research and conservation efforts in the EU habitats 2110 and 2120, and it highlights the importance of interdisciplinary collaboration in the field of habitat monitoring. This paper serves as a comprehensive description of the dataset and the methods used to collect it, making it a valuable resource for the scientific community.
The surveying of European Union (EU) Annex I habitat "8110 - Siliceous scree of the montane to snow levels (Androsacetalia alpinae and Galeopsietalia ladani)" is generally executed by humans. ...However, robots could increase human monitoring capabilities. To this end, we collected information on this habitat employing the quadrupedal robot ANYmal C. These data include videos of eight different typical or early warning species. Additionally, data on four relevés are provided. These consist, for instance, of the robot state, and videos and pictures collected to evaluate the habitat conservation status. The aim of this dataset is to help researchers in a variety of fields. For instance, information on plant species collected by the robot can be utilized to develop new procedures and new metrics to assess the habitat conservation status or to train neural networks for plant classification. On the other hand, engineers can use robot state information to validate their algorithms. This database is publicly available in the provided Zenodo repository.
The trend of current robotic research is to develop mobile robots that can perform highly dynamic tasks, which include jumping and running. To be employed profitably, this research necessitates a ...significant amount of work in the development of innovative planning and control algorithms that need experimental validation on actual robots. However, the majority of robots with highly dynamic performance capabilities are currently restricted to a few expensive platforms. This is a major obstacle that ultimately restricts the amount of contributors and the advancement of the research. Thus, a cost-effective actuator solution is needed that is also able to execute very dynamic movements. With this goal in mind, we present EM-Act, a modular series elastic actuator (SEA) for legged and multimodal dynamic robots. This work focuses on the development of the actuator solution by defining jump height as the prerequisite, identifying actuator parameters through simulations, and selecting and testing mechatronic elements of the design. The work also discusses a compact integration of desired compliance to address the impact forces. Furthermore, the work also details the implementation of the actuator solution on a 2-degree-of-freedom (DOF) robotic leg and experimentally validates its jumping performance.
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.
In this paper, we study the trajectory tracking problem using iterative learning control for continuous-time nonlinear systems with a generic fixed relative degree in the presence of disturbances. ...This class of controllers iteratively refine the control input relying on the tracking error of the previous trials and some properly tuned learning gains. Sufficient conditions on these gains guarantee the monotonic convergence of the iterative process. However, the choice of the gains is heuristically hand-tuned given an approximated system model and no information on the disturbances. Thus, in the cases of inaccurate knowledge of the model or iteration-varying measurement errors, external disturbances, and delays, the convergence condition is unlikely to be verified at every iteration. To overcome this issue, we propose a robust convergence condition, which ensures the applicability of the pure feedforward control even if other classical conditions are not fulfilled for some trials due to the presence of disturbances. Furthermore, we quantify the upper bound of the nonrepetitive disturbance that the iterative algorithm is able to handle. Finally, we validate the convergence condition simulating the dynamics of a two degrees of freedom underactuated arm with elastic joints, where one is active, and the other is passive, and a Franka Emika Panda manipulator.