Perception and Action in Complex Environments (PACE).
Reference number 642961 call: H2020-MSCA-ITN-2014
PACE is a large European ITN funded project. The PACE research and training programme sits at the interface between basic science, technology and clinics, in order to unveil how humans control and adapt their movements in complex, naturalistic environments. Such a research agenda has major consequences for understanding how these movements are impacted by specific brain insults and how these impairments can be compensated for via new rehabilitation methods. Improving rehabilitation programmes for sensory and motor disabilities across the lifespan is a major societal challenge in western countries and many obstacles need to be overcome. To provide but one example, with regard to eye-hand coordination of upper limb movement remaining abilities are rarely assessed in stroke patients or sensory-disabled children and this impacts both prognostic estimation and rehabilitation. New technologies, such as robotics or virtual reality, offer exciting opportunities in the perspective to transfer state-of-the-art knowledge from basic research on sensorimotor transformation into the clinical domain. To meet these societal challenges, it is crucial to train a new generation of early-stage researchers in a programme such as PACE where fundamental and applied/clinical research are effectively integrated via collaborative research, doctoral secondments and theoretical courses – in other words, one in which clinicians, neuroscientists, theoreticians and engineers can contribute around a well-defined problem: how humans acquire, lose and recover movement performance.
PACE is coordinated by the CRNS (Institute de la Timone, Marseille), and Joan López-Moliner is the PI of the research project to be carried out in Barcelona
Enhancement of sensory prediction and motor timing (PREDICT).
Reference number PSI2013-41568-P.
First, we will here study the optimization of predictive mechanisms based on optical information and use mainly interceptive tasks as a paradigm, although generalization of basic principles to a diversity of precision tasks would easily be made. Prediction involves some uncertainty of immediate future states due to transmission noise and ambiguity of sensory signals (optical variables in our case). One basic aim is to explore how to enhance the use of these signals in terms of reducing spatial and temporal errors by promoting robust 3D interpretations that can frame these sensory signals. Therefore we expect to find methods to improve temporal execution by resorting to people building 3D models of the scene. Second, predictive mechanisms appears to help humans cope with natural delays. Operating devices with additional delays between motor commands and expected feedback is becoming trendy in situations requiring high precision demands (tele-operations, drones, etc.). We easily adapt to these additional delays but at the cost of loosing precision. We try to answer which is the error-signal that drives this adaptation. This knowledge could be relevant for optimizing temporal performance under a diversity of situations in which additional delays are present. Finally, there is a strong neurocomputational perspective in which behaviourally revealed mechanisms will be implemented in neural simulations.
PREDICT's PIs are Joan López-Moliner and Matthias S. Keil
Collision detection in arbitrary environments.
TARGET, a Versatile Algorithm for Vision-based Detection of Object Approaches and Collision Threads. Ref. AVCRI-208
Matthias has developed the computer algorithm TARGET for predicting collision threads. TARGET processes image frames from video sequences or video cameras with no constraints on frame rate or resolution. The output is a simple signal which increases before collision would occur. TARGET suppresses background movement (as it would occur during car driving) without sacrificing its sensitivity for detecting collisions. Furthermore, large-field motion patterns due to camera movement or self-motion are also suppressed. Both suppression mechanisms contribute to a reduction in the number of false alerts, and make the algorithm versatile such that it can be used for many different applications, for example car driving, or airplanes. The algorithm is based on local computations, and a corresponding implementation in hardware should be feasible.