HIGHLY NEURAL OLFACTORIMOTOR ROBOTICS

While much of robotics focuses on visuomotor control, olfaction is the fundamental basis for sensorimotor control in the simpler creatures in which cortex first evolved.

Evolved Machines, a Palo Alto research group (www.evolvedmachines.com), aims to develop an artificial olfactory system that combines a high-dimension reversible olfactory sensor array of nanotube field-effect transistors with signal processing capability developed by reverse-engineering the olfactory circuitry of the brain.

This presentation will describe both elements of the system—the sensor and the neural back-end—and plans for combining these elements to enable autonomous robotic devices.

Evolved Machines is now using artificial evolution in virtual worlds that include odorant plumes to develop olfactorimotor algorithms for implementation in autonomous robotic devices that can find sources in the presence of the complex background conditions characteristic of real-world environments. It will be argued that because cortical visual systems evolved from the simpler olfactory cortex, the right place to start in the next generation of neural robotics may be olfactorimotor control.

Dr. Rhodes leads Evolved Machines, which is developing sensory device platforms based on the reverse engineering of biological neural circuitry. He also is serving as principal investigator and manager of two DARPA Defense Sciences Office teams. Dr. Rhodes has an AB in physics from Harvard University, an MS in physics from Stanford University, and a PhD in neuroscience from New York University Medical School. He has published more than 30 papers and abstracts.