The development of a new generation of aerial service robots capable to support human beings in all those activities which require the ability to interact actively and safely with environments not constrained on ground but, indeed, freely in air (e.g. inspection of buildings and large infrastructures, sample picking, aerial remote manipulation, etc.) is an open challenging field. The research is focused on the development of advanced automatic control algorithms able to govern the aerial platform which will be remotely supervised by the operator with the use of haptic devices. Particular emphasis is given to develop advanced human-in-the-loop and autonomous navigation control strategies relying upon a cooperative and adaptive interaction between the on-board automatic control and the remote operator. Force and visual feedback strategies are also investigated in order to transform the aerial platform in a "flying hand" suitable for aerial manipulation.
Biologically inspired robotic systems are becoming increasingly popular, especially in the field of medical robotics. The realization of robotic devices able to replicate the human behavior guarantees obtaining motor recovery, functional substitution and human-robot interaction as human-like as possible. From the study of the anatomy of the human hand and its behavior during grasping, it is possible to develop human-like grasping algorithms aimed at acquiring a better knowledge of the hand kinematics. This is useful to design new human-like robotic hands, new rehabilitation devices and to identify reference grasping strategies for understanding differences between healthy and pathological subjects. In assistive robotics, as well as in the field of hand prostheses, the ability of performing smooth movements and obtaining a stable grasp is essential. A bio-inspired approach for posture prediction and finger trajectory planning with a robotic hand has been developed, which appears to be rather promising. In order to support home-based rehabilitation, hand pose estimation algorithms have been developed and tested with a cheap camera system, such as the Kinect, and a system for patient performance evaluation during an ADL-based rehabilitation has been proposed.
COGNITIVE CONTROL OF ROBOTIC SYSTEMS
In order for a robot to perform complex tasks in the real work, it is necessary to endow its control system with cognitive capabilities. The processing architecture provides an intelligent behavior allowing the robot to learn and react to the world. Different approaches have been proposed taking inspiration from both animal and human behaviors. The adoption of an attentional mechanism is advantageous in terms of both efficient resources allocation and sensory-motor coordination. Several applications have been developed, showing a performance improvement in performance of the whole robotic system.
Dexterous manipulation of objects with two robot hands require a suitable cooperation between the robots. Control of the absolute motion of the object as well as of the internal forces shall be ensured. Also, the motions of the arms should be coordinated with those of the hands. In this framework, research is focused on design of effective control strategies that could be easily parameterised so as to preserve smoothness during the transitions at the contact with objects. Experimental validation is in course.
Most robot tasks imply interaction with the environment. Proper execution of constrained motion can be achieved using compliant control systems to accommodate external forces. A number of force control schemes have been developed, such as impedance control and parallel/force position control, where special attention has been devoted to the description of the end-effector orientation. Stability and robustness of such schemes has been analyzed and extensive testing on the experimental setup has been carried out.
Robots are becoming ubiquitous and utilized in anthropic environments. When a robot cohabits the environment with a human being, human-robot interaction issues are to be considered. It is customary to distinguish between cognitive and physical human-robot interaction. Suitable modelling and control techniques for collision avoidance and, more in general, for fast reactive strategies, are being devised, allowing a safe and dependable interaction. Experiments are developed on a humanoid manipulator available at DLR (German Aerospace Agency).
LIGHTWEIGHT FLEXIBLE ARMS
The adoption of lightweight materials in the realization of robot arms may provide advantages in terms of operational speeds and structural compliance. The price to pay is the complexity of the dynamic model and thus of the control techniques needed. Several controllers are being developed which account for the flexibility either concentrated at the joints or distributed in the links. The inclusion of force control features has also been studied.
MOBILE DYNAMIC NON-PREHENSILE MANIPULATION
With the increase of powerful technology in both sensing and actuation speed, it has become possible to manipulate an object in a very fast way. However, it is difficult for a robot to replicate human dexterous capabilities. By exploiting the dynamic effects of both object and robot motion, the robotic system can effectively achieve object manipulation in the desired way. In dynamic manipulation a relevant role is played by forces and accelerations, which are used together with kinematics, static and quasi-static forces, to achieve a general description of a manipulation task. Moreover, the non-prehensile manipulation of an object extends its feasible movements allowing rolling, pushing, throwing and tossing, and exhibits the intrinsic dynamics of both the task and the object. Hence, dynamic manipulation is a method which actively uses task dynamics to control motion variables of a given object. Performing such dynamic non-prehensile manipulation tasks by using deformable objects and mobile platforms is a great challenge.
MOTION GENERATION AND BIOMECHANICAL ANALYSIS FOR HUMAN FIGURES
Over the years manufacturing companies have considered man adaptability as a basic parameter of quality for their products (user-centered design), and thus an increasing attention has been devoted to ergonomic analyses, even from the early stage of design. In particular, Virtual Prototyping and Digital Human Modeling (DHM) are now advanced enough to correctly predict human-product and human–process interaction. Yet, digital human simulation still is a very time-consuming task, mainly because of the difficulty in controlling the kinematic chain of virtual humanoids by means of common key-frame based animation techniques. Some work has been conducted in developing algorithms capable of speeding up the posturing of digital human figures as needed.
MULTIFUNCTIONAL ROBOTIC HANDS: DESIGN AND CONTROL
This research is devoted to the development and the employment of artificial hands in all those fields where the use of anthropomorphic prehensile devices makes the difference, such as humanoid robots, prosthetics, surgical tools. The innovation will be linked not only to the mechanical design and integrated sensor systems but also to novel planning and control algorithms generating intelligent and effective actions. For the hand design, research starts from underactuated design solutions based on the concept of postural synergies and goes toward innovative solutions. On one hand, these are aimed at simplification of the manipulation device itself, for improving its affordability, in terms of weight and size, without reducing the functionalities. On the other, they are aimed at simplification of the control. About control and planning algorithms, one objective is that of merging vision and touch information and integrating them in reasoning to execute complex manipulation tasks. Furthermore, methods to learn from human data and to map synergies to artificial hands are investigated. Learning and control strategies based on neural networks, reinforcement learning and sensory-motor synergies are investigated and tested on anthropomorphic robotic hands.
ROBOTIC SURGERY TECHNOLOGY
Surgical robotics is a rapidly evolving field which helps to minimize the pain and risk associated with surgery, while increasing the likelihood of a fast recovery and excellent clinical outcomes. The introduction of a surgical robot into the operating room combines technological and clinical breakthroughs in developing new surgical techniques and approaches to improve the quality and outcome of the surgery. A simulator and a manipulator useful for surgery have been designed and developed: the preliminary experiments have been conducted on real patients and some promising results have been achieved.
SEMANTICS IN MEDICAL ROBOTICS
Medical robotics includes a number of devices used for surgery, medical training, rehabilitation therapy, prosthetics, and assisting people with disabilities. Nowadays, robotic devices are used to replace missing limbs, perform delicate surgical procedures, deliver neurorehabilitation therapy to stroke patients, teach children with learning disabilities, and perform a growing number of other health-related tasks. However, usually concrete models and specific knowledge are not available for objects or events in the medical robot's work environment. Hence, a medical robot system has to rely on more generalized modes of inference to infer the semantic content of the context. Adequate models and knowledge may describe broad categories of objects or events, acquired through training on sets of numerous examples. Knowledge may also be inferred from similarities and correspondences discovered between novel and known cases. As a matter of fact, there is a growing tendency to introduce high-level semantic knowledge into robotic systems. When a medical robot encounters unknown objects in its environment and semantic models are available, the perceptual system can derive knowledge from the relationships established with known objects of a similar typology. Moreover, through semantic modeling of low-level features within a scenario, robots can generate representation of such features at a level of abstraction where logical reasoning methods could be applied for decision making. Furthermore, at such semantic level more than one modality can be merged to complement each other and produce logical inferences. As a result, different cognitive systems have become quite popular among the research community, especially those using deep learning techniques over images and language sources, showing promising results.
According to a study of the International Federation of Robotics, in twenty years from now two thirds of the world robotics market will be covered by service robots. Differently from industrial robots, used mainly in manufacturing, the applications of service robots can be found in home, public sectors, medical welfare, bio-industrial scenarios. The common denominator is the degree of autonomy of the system. Some work on a fire-fighting robotic system for rescue and servicing in tunnels has been carried out.
The use of visual sensors for industrial and research applications have had a strong impulse in the latest years. The major motivations behind this scenario are the progressive cost reduction of high-performance vision systems as well as the capability to extract multiple information from a workspace in a non-invasive manner. Efficient multiple visual tracking of 3D objects in unstructured environments, visual servoing of industrial robots and object grasping are the focus of research.