Tracking computer mouse movement for studying visual information processing: applications to driving situations in reduced visibility conditions
Résumé
In order to propose sustainable solutions for mobility in the future, driver assistance and autonomous systems are under development. Driving is mostly a vision-based task involving complex cognitive processing in order to adapt to driving environment and driving conditions which may change rapidly. Thus, either for interacting with human drivers or simply to get inspiration from biological systems, understanding the mechanisms involved in the human visual system is beneficial. The long-term objective is to perform as well as the human being (or even better), in critical situations where rapid search of information in road scenes is decisive for safe mobility. Some critical situations could be generated by conditions of reduced visibility (night, fog, rain, glare), or complex environment including a profusion of objects of interest as well as distractors. Moreover critical objects such as obstacles or other road users –like pedestrians– may be hidden (partially or fully) and yet must be detected on time to avoid accident. For visual systems, either human or artificial, this variability introduces uncertainty about the composition of the scene. This system is therefore required to be highly adaptive and the human visual system offers solutions regarding this. This paper is dedicated to a psychophysical method to study human visual processing, which can be applied in the context of driving with reduced visibility conditions. Psychophysical measures such as response times or eye tracking have long been used to study visual behaviors and to infer neurocognitive models. An often reported limitation of the response time measurements, is that they do not provide much information about the underlying processes leading to the response. For instance, they make it difficult to distinguish between visual exploration, abstract cognition and motor processes. Nevertheless each contribute to the progressive accumulation and manipulation of information, and therefore partially determine the instant at which a response button would be pressed. Tracking computer mouse movements can bring some more information about these processes when used in a specific perceptual decision-making paradigm (Freeman, 2010). The trajectory of the mouse indeed reflects hesitations as well as the temporal course of the decision, as hand movements and decision making are spontaneously performed simultaneously. By controlling and providing -more or less- ambiguous visual stimuli, underlying cognitive processes can be inferred more precisely. We propose to make a review about this paradigm and to present applications to visual cognition research, and driving situations such as vision in fog conditions for either human or artificial models of vision.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |