# A Motion Taxonomy for Manipulation Embedding

arXiv: https://arxiv.org/abs/2007.06695 · HTML: https://mlsanity.com/paper/2007.06695/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Jul 2020.

## Built on by (most cited first, 9 of 9)

- [Continual Learning of Knowledge Graph Embeddings](https://mlsanity.com/paper/2101.05850/) (2021)
- [Bringing motion taxonomies to continuous domains via GPLVM on hyperbolic manifolds](https://mlsanity.com/paper/2210.01672/) (2022)
- [Towards Robust One-shot Task Execution using Knowledge Graph Embeddings](https://mlsanity.com/paper/2105.04484/) (2021)
- [Consolidating Trees of Robotic Plans Generated Using Large Language Models to Improve Reliability](https://mlsanity.com/paper/2401.07868/) (2024)
- [Lemmanaid: Neuro-Symbolic Lemma Conjecturing](https://mlsanity.com/paper/2504.04942/) (2025)
- [Ground Manipulator Primitive Tasks to Executable Actions using Large Language Models](https://mlsanity.com/paper/2308.06810/) (2023)
- [ATOM-Bench: A Real-World Benchmark for Atomic Skills and Compositional Generalization in Manipulation Policies](https://mlsanity.com/paper/2606.16826/) (2026)
- [Skill Generalization with Verbs](https://mlsanity.com/paper/2410.14118/) (2024)
- [Taxonomy-aware Dynamic Motion Generation on Hyperbolic Manifolds](https://mlsanity.com/paper/2509.21281/) (2025)

## Abstract (arXiv)

To represent motions from a mechanical point of view, this paper explores motion embedding using the motion taxonomy. With this taxonomy, manipulations can be described and represented as binary strings called motion codes. Motion codes capture mechanical properties, such as contact type and trajectory, that should be used to define suitable distance metrics between motions or loss functions for deep learning and reinforcement learning. Motion codes can also be used to consolidate aliases or cluster motion types that share similar properties. Using existing data sets as a reference, we discuss how motion codes can be created and assigned to actions that are commonly seen in activities of daily living based on intuition as well as real data. Motion codes are compared to vectors from pre-trained Word2Vec models, and we show that motion codes maintain distances that closely match the reality of manipulation.