# Difference between revisions of "Bootstrapping bilinear models of robotic sensorimotor cascades"

(htdb2wiki: creating page for 2010o_cm11-icra.html) |
(htdb2wiki: creating page for 2010g_cm10-wafr.html) |
||

Line 1: | Line 1: | ||

{{HTDB paper | {{HTDB paper | ||

| authors = Andrea Censi | | authors = Andrea Censi, Richard M Murray | ||

| title = Bootstrapping bilinear models of robotic sensorimotor cascades | | title = Bootstrapping bilinear models of robotic sensorimotor cascades | ||

| source = | | source = Workshop on the Algorithmic Foundations of Robotics (WAFR), 2010 (Submitted) | ||

| year = 2010 | | year = 2010 | ||

| type = | | type = Preprint | ||

| funding = | | funding = | ||

| url = http://www.cds.caltech.edu/~murray/preprints/ | | url = http://www.cds.caltech.edu/~murray/preprints/cm10-wafr_s.pdf | ||

| abstract = | | abstract = We consider the bootstrapping problem, which consists in learning a model of the agent's sensors and actuators starting from zero prior informa- tion, and we take the problem of servoing as a cross-modal task to validate the learned models. We study the class of sensors with bilinear dynamics, for which the derivative of the observations is a bilinear form of the control commands and the observations themselves. This class of models is simple, yet general enough to represent the main phenomena of three representative sensors (field sampler, camera, and range-finder), apparently very different from one another. It also allows a bootstrapping algorithm based on Hebbian learning, and a simple bioplausible control strategy. The convergence proper- ties of learning and control are demonstrated with extensive simulations and by analytical arguments. | ||

We consider the bootstrapping problem, which | |||

| flags = | | flags = | ||

| tag = | | tag = cm10-wafr | ||

| id = | | id = 2010g | ||

}} | }} |

## Latest revision as of 06:16, 15 May 2016

###
Andrea Censi, Richard M Murray

Workshop on the Algorithmic Foundations of Robotics (WAFR), 2010 (Submitted)

We consider the bootstrapping problem, which consists in learning a model of the agent's sensors and actuators starting from zero prior informa- tion, and we take the problem of servoing as a cross-modal task to validate the learned models. We study the class of sensors with bilinear dynamics, for which the derivative of the observations is a bilinear form of the control commands and the observations themselves. This class of models is simple, yet general enough to represent the main phenomena of three representative sensors (field sampler, camera, and range-finder), apparently very different from one another. It also allows a bootstrapping algorithm based on Hebbian learning, and a simple bioplausible control strategy. The convergence proper- ties of learning and control are demonstrated with extensive simulations and by analytical arguments.

- Preprint: http://www.cds.caltech.edu/~murray/preprints/cm10-wafr s.pdf
- Project(s):