José de Gea Fernández

Purpose of Review: The goal of this paper is to review current developments in the area of underwater robotics regarding the use of AI, especially in model learning, robot control, perception and navigation as well as manipulation.Recent Findings: AI technologies and advanced control techniques are finding their way into robotics systems to deal with complex and challenging conditions and to equip them with higher levels of autonomy.Summary: Although AI techniques and concepts are already a focus area in research on autonomous underwater systems, broad adoption to commercial systems is still in its infancy. Nonetheless, major advances have been done in recent years, especially on integrating different capabilities (perception, navigation, advanced control) in a single system and with first approaches on interaction and autonomous manipulation.

Last decades of humanoid research has shown that humanoids developed for high dynamic performance require a stiff structure and optimal distribution of mass~ inertial properties. Humanoid robots built with a purely tree type architecture tend to be bulky and usually suffer from velocity and force/torque limitations. This paper presents a novel series-parallel hybrid humanoid called RH5 which is 2 m tall and weighs only 62.5 kg capable of performing heavy-duty dynamic tasks with 5 kg payloads in each hand. The analysis and control of this humanoid is performed with whole-body trajectory optimization technique based on differential dynamic programming (DDP). Additionally, we present an improved contact stability soft-constrained DDP algorithm which is able to generate physically consistent walking trajectories for the humanoid that can be tracked via a simple PD position control in a physics simulator. Finally, we showcase preliminary experimental results on the RH5 humanoid robot.

Driven by advances in miniaturized electronics, many robotic systems today consist of modular hardware components. This leads to numerous computing units that are distributed within such systems. In order to make better use of such hardware structures from a computational point of view, the implementation of classical control approaches should be reconsidered. Following this idea, a method for distributed computation of motion dynamics of robotic systems using Field Programmable Gate Arrays is discussed. In this approach, local low-level actuator controllers are regarded as interconnected nodes, which are aware of their actuated degree of freedom and the resulting motion of the attached rigid body link element. A modified recursive Newton–Euler algorithm is computed by these nodes, where each node only exchanges data with the neighboring nodes. In the computations, a linear dependency on the dynamic parameters is kept in order to simplify the direct use of dynamic parameters estimated from experimental data. Implementation details and experimental results using a robotic manipulator arm are presented. The experiments show that the method allows compliant motion control. Payloads and external forces acting on a link are considered in the distributed computation of the model by informing the node controlling the respective link of the additional forces.

In this paper, we present a software-based approach for collision avoidance that can be applied in human–robot collaboration scenarios. One of the contributions is a method for converting clustered 3D sensor data into computationally efficient convex hull representations used for robot-obstacle distance computation. Based on the computed distance vectors, we generate collision avoidance motions using a potential field approach and integrate them with other simultaneously running robot tasks in a constraint-based control framework. In order to improve control performance, we apply evolutionary techniques for parameter optimization within this framework based on selected quality criteria. Experiments are performed on a dual-arm robotic system equipped with several depth cameras. The approach is able to generate task-compliant avoidance motions in dynamic environments with high performance.

This paper describes data-driven modelling methods and their use for the control of a novel set of series-elastic actuators (SEAs). A set of elastic actuators was developed in order to fulfill the end-user needs for tailored industrial collaborative robot manipulators of different morphologies and payloads. Three different types of elastic actuation were investigated, namely, disc springs, coil springs and torsion bars. The developed algorithms were validated both on single actuators and on a 6-DOF robotic arm composed of such actuators.

Continuum robots offer compliant and dexterous operations, which are suitable to be used in unstructured environments. Tendon-driven catheters, owing to their continuum structure, are applied in minimal invasive surgeries such as intracardiac interventions. However, due to the intrinsic nonlinearities and external disturbances, it is still a challenging task to accurately steer the catheter tip to the desired 3D positions. In this article, we proposed a new probabilistic kinematic model and a model-based three-dimensional position control scheme for a tendon-driven cardiac catheter. A dynamic Gaussian-based probabilistic model is developed to learn a mapping from the catheter states to the control actions. Based on the probabilistic model, a closed-loop position control is developed, in which the catheter is driven by a newly designed catheter driver system and tracked by a multiple near-infrared camera system. The proposed catheter framework is evaluated by the 3D trajectory tracking experiments both in a real 3D open space and in a minimum-energy-based simulator. The proposed control framework approximates the general kinematic by a combination of a catheter translation model and a distal workspace model, which provides the ability of automatically positioning the catheter tip in 3D and improving the accuracy by compensating the learned nonlinear effects.