Control Methodologies for Assistive Robots Operated via Brain Computer Interface

In the last decade, more than 700 million people are affected by disability or handicap. In this context, the research interest in assistive robotics is growing up. For people with mobility impairments daily life operations, as dressing or feeding, require the assistance of dedicated people; thus, the use of devices providing independent mobility can have a large impact on improving their life quality. Indeed, in the recent years, the research community is devoting a great effort towards the development of suitable robotic solutions to assist people with physical disabilities, also thanks to factors such as the availability of proper devices with lower prices and higher performance and robustness. From a robotic technology perspective, the specific assistive scenario can be characterized based on both the robotic system used to support the impaired person, and the Human-Machine Interface (HMI) adopted to allow the user to generate commands for the system. In particular, the HMI can rely on different types of signals, such as ElectroOculoGraphic (EOG), ElectroEncephaloGraphic (EEG) and ElectroMyoGraphic (EMG), and diversified HMIs have been developed in the latest years trying to find suitable solutions for specific applications and impairments. From the user perspective, the operation mode of such systems may depend on the level of autonomy provided by/required to the robotic system; from the device perspective, this correspond to different control modes that vary from shared to supervisory control. In shared mode the user is involved in the control loop of the system by continuously generating high-frequency motion commands; such commands are then translated from the control software in low-level functions after applying all the safety policies. In supervisory mode the user provides high level low-frequency commands (e.g., to start/stop actions) and the system operates in complete autonomy; the control software must generate motion directives that realize the required action while taking into account safety, comfort and efficiency. The operation modes of the robotic devices are strictly connected to the HMI used to generate and communicate commands. Among the different HMIs, Brain-Computer-Interfaces (BCIs) represent a relatively new technology that has recently attracted large attention in view of the fact that BCIs may be used in the absence of motion capability of the user with applications in different areas of assistive technologies as motor recovery, entertainment, communication and control. Indeed, BCIs have been recently proposed to drive wheelchairs, to guide robots for telepresence, to control exoskeletons and mobile robots. Most BCIs rely on non-invasive EEG signals, i.e. the electrical brain activity recorded from electrodes placed on the scalp. By processing such signals, the BCIs may allow the generation of commands that can be used for the communication with a software interface. EEG-BCI can be categorized based on the considered brain activity patterns, i.e.: Event-Related Desynchronization/Synchronization; Steady State Visual Evoke Potentials (SSVEP); P300 component of the Event Related Potentials (ERP); motor imagery. This thesis work focuses on the design of assistive robotics architectures to assist people with severe motor impairments in daily life activities, experimenting different methodologies and approaches to provide them a certain level of autonomy in different scenarios. The main contribution of this thesis work consists in the integration of BCIs paradigms and control architectures that allows to control systems with a large number of Degree Of Freedom (DOFs), autonomously taking into account low-level safety tasks both at joint and operational space levels. Controlling such a redundant system with a BCI would be indeed an overwhelming operation for users, thus the system has to own a certain level of autonomous capabilities. In particular, redundant systems can be exploited to perform multiple tasks simultaneously. For this reason, the Multi-Task Priority (MTP) inverse kinematics algorithm can be used. As well known in literature, the latter is based on the Closed Loop Inverse Kinematics (CLIK) and allows to manage a prioritized hierarchy of equality-based tasks, which are control functions characterized by a specific desired value, e.g., position and orientation. More recently, the MTP has been extended to handle also set-based tasks, namely Set-Based Task-Priority Inverse Kinematics (SBTPIK), in which the task value has to be kept in a certain set of values, i.e., below an upper threshold and above a lower one. This is the case of tasks such as the mechanical joint limits for a manipulator or the obstacle avoidance for a mobile robot. Therefore, it has been necessary to manage multiple tasks ensuring the safety system. Thus, the SBTPIK has been applied to fulfil this objective. Moreover, this thesis work focuses its attention on the developing of assistive robotics architecture with multi-robot systems, since they allow to accomplish complex tasks otherwise impossible for a single unit. Common control approaches for multi-robot systems are based on distributed architectures, where each robot computes its own control input only based on local information from onboard sensors or received form its neighbor robots. This means that the failure of one or more agents might jeopardize the task execution. For this reason, Fault Detection and Isolation (FDI) strategies become crucial so as to accomplish the assigned task also in the aforementioned case. In this context, a distributed Fault Diagnosis architecture aimed at detecting failures in a team of robots working in tight cooperation has been devised. The proposed approach relies on an distributed observer-controller scheme, where each robot estimates the overall system state by means of a local observer, and it uses such as an estimate to compute the local control input to achieve a specific task. The local observer is also used to define a set of residual vectors aimed to detect and isolate faults occurring on any robot of the team, even there is not direct communication. The research activities involving BCIs to control robots, have been intensified with a six months research period abroad at the Robotics and Mechatronics Center (RMC) of the Deutsches Zentrum für Luft-und Raumfahrt (DLR), located in Munich (Germany). In particular, the following activities were conducted: the integration of an EEG system with the hardware and software architecture of the robotic platform EDAN; the development of an EEG-based interface to command EDAN (EMG-controlled Daily Assistant); the evaluation of the EEG-based interface in a small user-study.