• Rural Digital Europe
• Publications close
• FI close
• NARCIS close

Local-HDP (for Local Hierarchical Dirichlet Process) is a hierarchical Bayesian method that has recently been used for open-ended 3D object category recognition. This method has been proven to be efficient in real-time robotic applications. However, the method is not robust to a high degree of occlusion. We address this limitation in two steps. First, we propose a novel semantic 3D object-parts segmentation method that has the flexibility of Local-HDP. This method is shown to be suitable for open-ended scenarios where the number of 3D objects or object parts is not fixed and can grow over time. We show that the proposed method has a higher percentage of mean intersection over union, using a smaller number of learning instances. Second, we integrate this technique with a recently introduced argumentation-based online incremental learning method, thereby enabling the model to handle a high degree of occlusion. We show that the resulting model produces an explicit set of explanations for the 3D object category recognition task. Accepted at ICRA 2023 Conference

Existing Deep Learning (DL) frameworks typically do not provide ready-to-use solutions for robotics, where very specific learning, reasoning, and embodiment problems exist. Their relatively steep learning curve and the different methodologies employed by DL compared to traditional approaches, along with the high complexity of DL models, which often leads to the need of employing specialized hardware accelerators, further increase the effort and cost needed to employ DL models in robotics. Also, most of the existing DL methods follow a static inference paradigm, as inherited by the traditional computer vision pipelines, ignoring active perception, which can be employed to actively interact with the environment in order to increase perception accuracy. In this paper, we present the Open Deep Learning Toolkit for Robotics (OpenDR). OpenDR aims at developing an open, non-proprietary, efficient, and modular toolkit that can be easily used by robotics companies and research institutions to efficiently develop and deploy AI and cognition technologies to robotics applications, providing a solid step towards addressing the aforementioned challenges. We also detail the design choices, along with an abstract interface that was created to overcome these challenges. This interface can describe various robotic tasks, spanning beyond traditional DL cognition and inference, as known by existing frameworks, incorporating openness, homogeneity and robotics-oriented perception e.g., through active perception, as its core design principles. Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public. Learning & Autonomous Control Joining and Mechanical Behavior

Several safe boundaries of critical Earth system processes have already been crossed due to human perturbations; not accounting for their interactions may further narrow the safe operating space for humanity. Using expert knowledge elicitation, we explored interactions among seven variables representing Earth system processes relevant to food production, identifying many interactions little explored in Earth system literature. We found that green water and land system change affect other Earth system processes strongly, while land, freshwater and ocean components of biosphere integrity are the most impacted by other Earth system processes, most notably blue water and biogeochemical flows. We also mapped a complex network of mechanisms mediating these interactions and created a future research prioritization scheme based on interaction strengths and existing knowledge gaps. Our study improves the understanding of Earth system interactions, with sustainability implications including improved Earth system modelling and more explicit biophysical limits for future food production. Determining the safe operating space for sustainable food production depends on the interactions of multiple processes within the Earth system. Expert knowledge provides critical insight into how these processes interact that improves Earth system modelling and our understanding of the limits of global food production. | openaire: EC/H2020/819202/EU//SOS.aquaterra Peer reviewed

A description and test results for a new method for automatically isolating individual trees from UAV LiDAR point clouds is presented. The isolation method is based on shortest path computations with height from the ground working as a restriction. The method is tested on a 4 ha tropical forest plot, which is also scanned with TLS to provide comparison data for the isolation results. The comparison results show that on average the intersection volume of the common voxel volume between TLS and UAV trees covers 56 % of the TLS tree and 44 % of the UAV trees. Peer reviewed

Demand for high spatial and temporal resolution measurements has triggered a rapid development of unmanned aircraft systems (UAS) for plant phenotyping and precision farming purposes. Similarly, recent progress in low-altitude remote sensing of solar-induced chlorophyll fluorescence (SIF) resulted in several studies aiming at the development of SIF proximal sensing approaches. Although first experimental results are promising, the requirements for reliable and repeatable measurements in agricultural experiments still constrain applicability of these platforms. In this study, we analyze current capabilities and potentials of SIF measuring UAS for operational use. We highlight existing challenges and outline how UAS SIF sensing could be used more frequently and reliably in precision agriculture applications in the near future. Peer reviewed

This work presents a novel fault-tolerant control scheme based on active inference. Specifically, a new formulation of active inference which, unlike previous solutions, provides unbiased state estimation and simplifies the definition of probabilistically robust thresholds for fault-tolerant control of robotic systems using the free-energy. The proposed solution makes use of the sensory prediction errors in the free-energy for the generation of residuals and thresholds for fault detection and isolation of sensory faults, and it does not require additional controllers for fault recovery. Results validating the benefits in a simulated 2-DOF manipulator are presented, and future directions to improve the current fault recovery approach are discussed. Comment: 7 pages, 6 figures, Accepted at the European Control Conference (ECC) 2021