Leveraging scene embeddings for gradient-based motion planning in latent space
Motion planning framed as optimisation in structured latent spaces has recently emerged as competitive with traditional methods in terms of planning success while significantly outperforming them in terms of computational speed. However, the real-world applicability of recent work in this domain rem...
| Autores principales: | , , , , |
|---|---|
| Formato: | Conference item |
| Lenguaje: | English |
| Publicado: |
IEEE
2023
|
| _version_ | 1826311726302232576 |
|---|---|
| author | Yamada, J Hung, CM Collins, J Havoutis, I Posner, I |
| author_facet | Yamada, J Hung, CM Collins, J Havoutis, I Posner, I |
| author_sort | Yamada, J |
| collection | OXFORD |
| description | Motion planning framed as optimisation in structured latent spaces has recently emerged as competitive with traditional methods in terms of planning success while significantly outperforming them in terms of computational speed. However, the real-world applicability of recent work in this domain remains limited by the need to express obstacle information directly in state-space, involving simple geometric primitives. In this work we address this challenge by leveraging learned scene embeddings together with a generative model of the robot manipulator to drive the optimisation process. In addition, we introduce an approach for efficient collision checking which directly regularises the optimisation undertaken for planning. Using simulated as well as real-world experiments, we demonstrate that our approach, AMP-LS, is able to successfully plan in novel, complex scenes while outperforming traditional planning baselines in terms of computation speed by an order of magnitude. We show that the resulting system is fast enough to enable closed-loop planning in real-world dynamic scenes. |
| first_indexed | 2024-03-07T08:15:30Z |
| format | Conference item |
| id | oxford-uuid:528ee1fe-e9dd-42e5-846a-1ec4feb6611d |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T08:15:30Z |
| publishDate | 2023 |
| publisher | IEEE |
| record_format | dspace |
| spelling | oxford-uuid:528ee1fe-e9dd-42e5-846a-1ec4feb6611d2023-12-20T14:49:17ZLeveraging scene embeddings for gradient-based motion planning in latent spaceConference itemhttp://purl.org/coar/resource_type/c_5794uuid:528ee1fe-e9dd-42e5-846a-1ec4feb6611dEnglishSymplectic ElementsIEEE2023Yamada, JHung, CMCollins, JHavoutis, IPosner, IMotion planning framed as optimisation in structured latent spaces has recently emerged as competitive with traditional methods in terms of planning success while significantly outperforming them in terms of computational speed. However, the real-world applicability of recent work in this domain remains limited by the need to express obstacle information directly in state-space, involving simple geometric primitives. In this work we address this challenge by leveraging learned scene embeddings together with a generative model of the robot manipulator to drive the optimisation process. In addition, we introduce an approach for efficient collision checking which directly regularises the optimisation undertaken for planning. Using simulated as well as real-world experiments, we demonstrate that our approach, AMP-LS, is able to successfully plan in novel, complex scenes while outperforming traditional planning baselines in terms of computation speed by an order of magnitude. We show that the resulting system is fast enough to enable closed-loop planning in real-world dynamic scenes. |
| spellingShingle | Yamada, J Hung, CM Collins, J Havoutis, I Posner, I Leveraging scene embeddings for gradient-based motion planning in latent space |
| title | Leveraging scene embeddings for gradient-based motion planning in latent space |
| title_full | Leveraging scene embeddings for gradient-based motion planning in latent space |
| title_fullStr | Leveraging scene embeddings for gradient-based motion planning in latent space |
| title_full_unstemmed | Leveraging scene embeddings for gradient-based motion planning in latent space |
| title_short | Leveraging scene embeddings for gradient-based motion planning in latent space |
| title_sort | leveraging scene embeddings for gradient based motion planning in latent space |
| work_keys_str_mv | AT yamadaj leveragingsceneembeddingsforgradientbasedmotionplanninginlatentspace AT hungcm leveragingsceneembeddingsforgradientbasedmotionplanninginlatentspace AT collinsj leveragingsceneembeddingsforgradientbasedmotionplanninginlatentspace AT havoutisi leveragingsceneembeddingsforgradientbasedmotionplanninginlatentspace AT posneri leveragingsceneembeddingsforgradientbasedmotionplanninginlatentspace |