Neurophysiological principles for technical implementations

  • Print
Details
Category: Homepage
Published on Tuesday, 16 October 2012 00:00
Written by dominique martinez
Hits: 8255

We apply neurophysiological principles to build invariant representations from sensory data.

  • Robotic source seeking in turbulent medium. Searching for a source releasing particles in a turbulent medium, e.g. CBRN materials, is particularily challenging. The sensory landscape is heterogeneous and unsteady and the searcher has to rely on intermittent binary cues. Insects can be very efficient at solving this problem. For robotic searches, we apply information-theoric strategies like Infotaxis as it has been shown to produce trajectories similar to insects, e.g. moths attracted by a sexual pheromone.
  • Latency coding in artificial olfaction. In the context of fast olfactory processing, relative latencies of different neurons may carry concentration-invariant information about stimulus identity. We designed simple yet robust latency coding schemes for processing gas sensor data (Chen et al., 2011; Yamani et al., 2012). The response of the gas sensor array is mapped into  latency patterns, whose rank order is concentration-invariant. This study pioneers the translation of neurophysiological findings into hardware for the processing of electronic noses.
  • Invariant representations and blind source separation. Perception results from the way sensory information is transformed in the brain. Whereas sensory input varies rapidly (e.g. pixel intensity across an input image), hidden perceptual parameters (e.g. surface depth) vary smoothly over time.We derived a non-linear technique for extracting slowly varying features from rapidly varying signals and for non-linear blind source separation (Martinez and Bray, 2003).

Relevant publications:

Martinez D., Burgués J., Marco S. (2019) Fast measurements with MOX sensors: A least-squares approach to blind deconvolution. Sensors.

Zhang S., Martinez D., Masson JB. (2015) Multi-robot searching with sparse binary cues and limited space perception. Frontiers in Robotics and AI.

Martin-Moraud E. and Martinez D. (2010) Effectiveness and robustness of robot infotaxis for searching in dilute conditions. Frontiers in neurorobotics.

Yamani J., Boussaid F., Bermak A., and Martinez D. (2012) Glomerular latency coding in artificial olfaction. Frontiers in Neuroengineering, 4:18. doi: 10.3389/fneng.2011.00018.

Chen H.T., Ng K.T., Bermak A., Law M.K. and Martinez D. (2011) Spike latency coding in a biologically inspired micro-electronic nose. IEEE Trans. Biomedical Circuits and Systems, 5:2, 160-168.

Martinez D. and Bray A. (2003) Nonlinear blind source separation using kernels. IEEE Transactions on Neural Networks, Vol 14, no 1, pp. 228-236.