7 January 2022 in IEEE Transactions on Network Science and Engineering
This paper addresses a distributed connectivity control problem in networked multi-agent systems. The system communication topology is controlled through the algebraic connectivity measure, the second smallest eigenvalue of the communication graph Laplacian. The algebraic connectivity is estimated locally in a decentralized manner through a trust-based consensus algorithm, in which the agents communicate the perceived quality of the communication links in the system with their set of neighbors. In the presented approach, link qualities represent the weights of the communication graph from which the adjacency matrix is estimated. The Laplacian matrix and its eigenvalues, including the algebraic connectivity, are then calculated from this local estimate of the global adjacency matrix. A method for network topology control is proposed, which creates and deletes communication links based on the Albert-Barabási probabilistic model, depending on the estimated and referenced connectivity level. The proposed algebraic connectivity estimation and connectivity maintenance strategy have been validated both in simulation and on a physical robot swarm, demonstrating the method performance under varying initial topology of the communication graph, different multi-agent system sizes, in various deployment scenarios, and in the case of agent failure.
10 July 2019 in IEEE Robotics and Automation Letters
A common approach in the field of tactile robotics is the development of a new perception algorithm for each new application of existing hardware solutions. In this letter, we present a method of dimensionality reduction of an optical-based tactile sensor image output using a convolutional neural network encoder structure. Instead of using various complex perception algorithms, and/or manually choosing task-specific data features, this unsupervised feature extraction method allows simultaneous online deployment of multiple simple perception algorithms on a common set of black-box features. The method is validated on a set of benchmarking use cases. Contact object shape, edge position, orientation, and indentation depth are estimated using shallow neural networks and machine learning models. Furthermore, a contact force estimator is trained, affirming that the extracted features contain sufficient information on both spatial and mechanical characteristics of the manipulated object.
1 December 2021 in Intelligent Service Robotics
25 February 2021 in IEEE Robotics and Automation Letters
This letter presents a compliant exploration framework based on a collaborative robot Franka Panda that builds a 3D plant stem model. The model is built for agricultural plant treatment procedures without external sensors, as contact forces are estimated from joint torques and robot's dynamic and kinematic model. By devising an impedance-based exploratory control algorithm capable of following an unknown shape, while being provided with only a general direction in which to explore, we eliminate the need for a precise position controller. Our approach is validated through experiments with several mock-up plant stems, showing that the proposed framework is capable of building a satisfactory 3D model of a plant. The method is evaluated against the ground truth model, and compared to the state of the art approach based on an industrial manipulator with external sensors.
1 March 2022 in Field Robotics
11 October 2018 in IFAC-PapersOnLine