Brain-computer interfaces, as any other interaction modality based on physiological signals and body channels (e.g., muscular activity, speech and gestures), are prone to errors in the recognition of subject’s intent. In this paper we exploit a unique feature of the “brain channel”, namely that it carries information about cognitive states that are crucial for a purposeful interaction. One of these states is the awareness of erroneous responses. Different physiological studies have shown the presence of error-related potentials (ErrP) in the EEG recorded right after people get aware they have made an error. However, for human-computer interaction, the central question is whether ErrP are also elicited when the error is made by the interface during the recognition of the subject’s intent and no longer by errors of the subject himself. In this paper we report experimental results with three volunteer subjects during a simple human-robot interaction (i.e., bringing the robot to either the left or right side of a room) that seem to reveal a new kind of ErrP, which is satisfactorily recognized in single trials. These recognition rates significantly improve the performance of the brain interface. 1

Abstract — Human control of high degree-of-freedom robotic systems is often difficult due to the overwhelming number of variables that need to be specified. Instead, we propose the use of sparse control subspaces embedded within the pose space of a robotic system. Using captured human motion for training, we address this sparse control problem by uncovering 2D subspaces that allow cursor control, or eventually decoding of neural activity, to drive a robotic hand. Considering the problems in previous work related to noise in pose graph construction and motion capture data, we introduce a method for denoising neighborhood graphs for embedding hand motion into 2D spaces. We present results demonstrating our approach to interactive sparse control for successful power grasping and precision grasping using a 13 DOF robot hand. I.

The promise of Brain-Computer Interfaces (BCI) technology is to augment human capabilities by enabling interaction with computers through a conscious and spontaneous modulation of the brainwaves after a short training period. Indeed, by analyzing brain electrical activity online, several groups have designed brain-actuated devices that provide alternative channels for communication, entertainment and control. Thus, a person can write messages using a virtual keyboard on a computer screen and also browse the internet. Alternatively, subjects can operate simple computer games, or brain games, and interact with educational software. Work with humans has shown that it is possible for them to move a cursor and even to drive a wheelchair. This paper briefly reviews the field of BCI, with a focus on non-invasive systems based on electroencephalogram (EEG) signals. It also describes three brain-actuated devices we have developed: a virtual keyboard, a brain game, and a wheelchair. Finally, it shortly discusses current research directions we are pursuing in order to improve the performance and robustness of our BCI system, especially for real-time control of brainactuated robots.

Abstract — Human control of high degree-of-freedom robotic systems, e.g. anthropomorphic robot hands, is often difficult due to the overwhelming number of variables that need to be specified. Previous work has addressed this sparse control problem by learning a high-dimensional manifold of robot poses to provide low-dimensional control subspaces. Such subspaces allow cursor control, or eventually decoding of neural activity, to drive a robotic hand. Considering previously identified problems related to noise in manifold learning, we introduce a method for denoising neighborhood graphs in order to embed hand motion into 2D spaces. We present results demonstrating our approach in the case of a synthetic swissroll as well as in the embeddings for interactive sparse control for several grasping tasks. I.

Summary. Neurally controlled prosthetic devices capable of environmental manipulation have much potential towards restoring the physical functionality of disabled people. However, the number of user input variables provided by current neural decoding systems is much less than the number of control degrees-of-freedom (DOFs) of a prosthetic hand and/or arm. To address this sparse control problem, we propose the use of low-dimensional subspaces embedded within the pose space of a robotic limb. These subspaces are extracted using dimension reduction techniques to compress captured human hand motion into a (often two-dimensional) subspace that can be spanned by the output of neural decoding systems. To evaluate our approach, we explore a set of current state-of-the-art dimension reduction techniques and show results for effective control of a 13 DOF robot hand performing basic grasping tasks taking place in both static and dynamic environments. 1

Because functional near-infrared spectroscopy (fNIRS) eases many of the restrictions of other brain sensors, it has potential to open up new possibilities for HCI research. From our experience using fNIRS technology for HCI, we identify several considerations and provide guidelines for using fNIRS in realistic HCI laboratory settings. We empirically examine whether typical human behavior (e.g. head and facial movement) or computer interaction (e.g. keyboard and mouse usage) interfere with brain measurement using fNIRS. Based on the results of our study, we establish which physical behaviors inherent in computer usage interfere with accurate fNIRS sensing of cognitive state information, which can be corrected in data analysis, and which are acceptable. With these findings, we hope to facilitate further adoption of fNIRS brain sensing technology in HCI research. ACM Classification: H5.2 [Information interfaces and

michael.tangermann}

Abstract — We investigate the use of five dimension reduction and manifold learning techniques to estimate a 2D subspace of hand poses for the purpose of generating motion. Our aim is to uncover a 2D parameterization from optical motion capture data that allows for transformation sparse user input trajectories into desired hand movements. The use of shape descriptors for representing hand pose is additionally explored for dealing with occluded parts of the hand during data collection. We present early results from uncovering 2D parameterizations of power and precision grasps and their use to drive a physically simulated hand from 2D mouse input. I.

Neurally controlled prosthetic devices capable of object manipulation have much potential towards restoring the physical functionality of disabled individuals. However, the number of user input variables provided by current neural decoding systems is much less than the number of control degrees-of-freedom (DOFs) of a prosthetic hand and/or arm. More specifically, efforts to decode user neural activity coming from feasible sensing technologies, such as electroencephalogram (EEG) [3] and cortical neural implants [4], [8], into control signals have demonstrated success limited to 2-3 DOFs with bandwidth approximately 15 bits/sec. for effective control of a DLR/HIT robot hand with 13 DOF. The grasping tasks performed contain finger tapping motions, powergrasps and precision grasps. Example tasks include grasping of a bottle of water, an eraser, an oscillating box and simulation of the grasping actions taking place while having a meal. This work is illustrated in more detail in

Human control of high degree-of-freedom robotic systems, e.g. anthropomorphic robot hands, is often difficult due to the overwhelming number of variables that need to be specified. The problem is magnified for applications to biorobotics where efforts to decode user neural activity into control signals have demonstrated success limited to 2-3 DOFs with bandwidth approximately 15 bits/sec [5]. We address this sparse control problem by learning a high-dimensional manifold of robot poses in order to provide 2D subspaces for interactive control of a high-DOF robot hand. tasks