RECORD
MimicAgent: Quadruped Skills via Prompt-to-Trajectory Generation
Robotics · Submitted 21 Sep 2026
abstract compiled · 2026-09-24 · no language-model review
The authors' main claim These coarse reference trajectories are then used to train example-guided RL policies that are deployable in simulation and in the real-world.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
We present MimicAgent, a prompt-to-trajectory generation framework for learning dynamic quadruped skills. Although reward shaping is extensively used when training quadruped policies, navigating the resulting reward landscape is notoriously difficult, requiring hours of "graduate student descent". Eureka attempts to automate reward design with LLMs, but we find that it struggles to generalize across diverse skills and morphologies. Our key observation is that it is far easier for a human - and by association, an LLM - to generate reference motions than to shape reward functions. Our hypothesis is motivated by the success of example-guided RL for humanoids, which exploits large-scale motion capture datasets as references for training locomotion policies. Unlike humanoids, quadrupeds lack such reference motion data. Towards this end, we propose MimicAgent, an agentic harness that, given a skill prompt, generates quadruped reference trajectories with coding agents. These coarse reference trajectories are then used to train example-guided RL policies that are deployable in simulation and in the real-world. Notably, we find that when prompting Claude Fable 5.1 within our agentic harness, 87% of prompts yield semantically aligned reference trajectories.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | We present MimicAgent, a prompt-to-trajectory generation framework for learning dynamic quadruped skills. | contribution 1.00 | 0.87 | 0.13 | asserted 1.00 |
| s1 | Although reward shaping is extensively used when training quadruped policies, navigating the resulting reward landscape is notoriously difficult, requiring hours of "graduate student descent". | background 0.99 | 0.09 | 0.03 | asserted 1.00 |
| s2 | Eureka attempts to automate reward design with LLMs, but we find that it struggles to generalize across diverse skills and morphologies. | background 0.59 | 0.85 | 0.04 | asserted 1.00 |
| s3 | Our key observation is that it is far easier for a human - and by association, an LLM - to generate reference motions than to shape reward functions. | interpretation 0.48 | 0.59 | 0.27 | asserted 1.00 |
| s4 | Our hypothesis is motivated by the success of example-guided RL for humanoids, which exploits large-scale motion capture datasets as references for training locomotion policies. | background 0.93 | 0.10 | 0.05 | asserted 0.92 |
| s5 | Unlike humanoids, quadrupeds lack such reference motion data. | gap 0.94 | 0.14 | 0.03 | asserted 1.00 |
| s6 | Towards this end, we propose MimicAgent, an agentic harness that, given a skill prompt, generates quadruped reference trajectories with coding agents. | method 0.53 | 0.74 | 0.13 | asserted 0.96 |
| s7 | These coarse reference trajectories are then used to train example-guided RL policies that are deployable in simulation and in the real-world. | method 0.95 | 0.83 | 0.19 | asserted 1.00 |
| s8 | Notably, we find that when prompting Claude Fable 5.1 within our agentic harness, 87% of prompts yield semantically aligned reference trajectories. | result 1.00 | 0.96 | 0.15 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
We present MimicAgent, a prompt-to-trajectory generation framework for learning dynamic quadruped skills. Although reward shaping is extensively used when training quadruped policies, navigating the resulting reward landscape is notoriously difficult, requiring hours of "graduate student descent". Eureka attempts to automate reward design with LLMs, but we find that it struggles to generalize across diverse skills and morphologies. Our key observation is that it is far easier for a human - and by association, an LLM - to generate reference motions than to shape reward functions. Our hypothesis is motivated by the success of example-guided RL for humanoids, which exploits large-scale motion capture datasets as references for training locomotion policies. Unlike humanoids, quadrupeds lack such reference motion data. Towards this end, we propose MimicAgent, an agentic harness that, given a skill prompt, generates quadruped reference trajectories with coding agents. These coarse reference trajectories are then used to train example-guided RL policies that are deployable in simulation and in the real-world. Notably, we find that when prompting Claude Fable 5.1 within our agentic harness, 87% of prompts yield semantically aligned reference trajectories.
Human review
References
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.