Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences
Robot kinematic data, despite being high-dimensional, is highly correlated, especially when considering motions grouped in certain primitives. These almost linear correlations within primitives allow us to interpret motions as points drawn close to a union of low-dimensional affine subspaces in the...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2022-09-01
|
| Series: | Frontiers in Neurorobotics |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fnbot.2022.891031/full |
| _version_ | 1811261156043522048 |
|---|---|
| author | Vsevolod Nikulin Jun Tani |
| author_facet | Vsevolod Nikulin Jun Tani |
| author_sort | Vsevolod Nikulin |
| collection | DOAJ |
| description | Robot kinematic data, despite being high-dimensional, is highly correlated, especially when considering motions grouped in certain primitives. These almost linear correlations within primitives allow us to interpret motions as points drawn close to a union of low-dimensional affine subspaces in the space of all motions. Motivated by results of embedding theory, in particular, generalizations of the Whitney embedding theorem, we show that random linear projection of motor sequences into low-dimensional space loses very little information about the structure of kinematic data. Projected points offer good initial estimates for values of latent variables in a generative model of robot sensory-motor behavior primitives. We conducted a series of experiments in which we trained a Recurrent Neural Network to generate sensory-motor sequences for a robotic manipulator with 9 degrees of freedom. Experimental results demonstrate substantial improvement in generalization abilities for unobserved samples during initialization of latent variables with a random linear projection of motor data over initialization with zero or random values. Moreover, latent space is well-structured such that samples belonging to different primitives are well separated from the onset of the training process. |
| first_indexed | 2024-04-12T18:58:13Z |
| format | Article |
| id | doaj.art-f8bcf251814348558474d8f4aa83af71 |
| institution | Directory Open Access Journal |
| issn | 1662-5218 |
| language | English |
| last_indexed | 2024-04-12T18:58:13Z |
| publishDate | 2022-09-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neurorobotics |
| spelling | doaj.art-f8bcf251814348558474d8f4aa83af712022-12-22T03:20:16ZengFrontiers Media S.A.Frontiers in Neurorobotics1662-52182022-09-011610.3389/fnbot.2022.891031891031Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequencesVsevolod NikulinJun TaniRobot kinematic data, despite being high-dimensional, is highly correlated, especially when considering motions grouped in certain primitives. These almost linear correlations within primitives allow us to interpret motions as points drawn close to a union of low-dimensional affine subspaces in the space of all motions. Motivated by results of embedding theory, in particular, generalizations of the Whitney embedding theorem, we show that random linear projection of motor sequences into low-dimensional space loses very little information about the structure of kinematic data. Projected points offer good initial estimates for values of latent variables in a generative model of robot sensory-motor behavior primitives. We conducted a series of experiments in which we trained a Recurrent Neural Network to generate sensory-motor sequences for a robotic manipulator with 9 degrees of freedom. Experimental results demonstrate substantial improvement in generalization abilities for unobserved samples during initialization of latent variables with a random linear projection of motor data over initialization with zero or random values. Moreover, latent space is well-structured such that samples belonging to different primitives are well separated from the onset of the training process.https://www.frontiersin.org/articles/10.3389/fnbot.2022.891031/fullgenerative modelsroboticslatent encodingrandom projectionmotion primitivesRecurrent Neural Network |
| spellingShingle | Vsevolod Nikulin Jun Tani Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences Frontiers in Neurorobotics generative models robotics latent encoding random projection motion primitives Recurrent Neural Network |
| title | Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences |
| title_full | Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences |
| title_fullStr | Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences |
| title_full_unstemmed | Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences |
| title_short | Initialization of latent space coordinates via random linear projections for learning robotic sensory-motor sequences |
| title_sort | initialization of latent space coordinates via random linear projections for learning robotic sensory motor sequences |
| topic | generative models robotics latent encoding random projection motion primitives Recurrent Neural Network |
| url | https://www.frontiersin.org/articles/10.3389/fnbot.2022.891031/full |
| work_keys_str_mv | AT vsevolodnikulin initializationoflatentspacecoordinatesviarandomlinearprojectionsforlearningroboticsensorymotorsequences AT juntani initializationoflatentspacecoordinatesviarandomlinearprojectionsforlearningroboticsensorymotorsequences |