Aero-VPA: Few-Step Flow-Matching Control over Non-Reconstructive JEPA Latents for Agile Flight on Edge Hardware
Aero-VPA: Few-Step Flow-Matching Control over Non-Reconstructive JEPA Latents for Agile Flight on Edge Hardware
This research introduces Aero-VPA, a lightweight Vision-Predictor-Action architecture designed for 3D navigation on edge hardware.
The paper presents Aero-VPA, a lightweight architecture for agile flight on edge hardware, validated via PX4 software-in-the-loop simulations in procedurally generated forests; it reports specific success rate improvements and crash reductions achieved by combining DAgger training with a map-free runtime shield, while noting that behavior cloning failed completely on the test set.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S4
Reported numbers
policy size
72.6
View original evidence
nts gained 30.0-34.4 percentage points, at mean speeds of 1.16-1.36 m/s. The code (https://github.com/ylee-sling/Aero-VPA), the trained policy (72.6 MB, https://huggingface.co/ylee-sling/aero-vpa) and the full benchmark dataset are public.
Open source S6
- code: Available Artifact link Open source S6
- dataset: Available Artifact link Open source S6
- paper: not_reported
- code: not_reported
- weights: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
out results. This version reports the first closed-loop evaluation: PX4 software-in-the-loop simulation of quadrotor missions in procedurally generated forests. Behaviour cloning failed on all 60 missions despite open-loop errors of 0.02-0.04 m/s: the policy copies its last command, a copycat form of causal confusion. DAgger with a privileged, safety-gated supervisor gave the first successes but plateaued at 11.1-21.
Open source S4
1% across six checkpoints. A map-free runtime shield with a 15 s depth memory, barrier constraints, a cap on motion outside the 73-degree field of view and backoff along the flown path raised the first DAgger policy's success on 180 paired missions from 21.1% to 52.2% (95% CI [45.0, 59.4]; exact McNemar p = 1.2 x 10^-10) and cut its crashes from 72.2% to 22.2%, while timeouts rose from 2.8% to 23.3%. All six checkpoi
Open source S5
nts gained 30.0-34.4 percentage points, at mean speeds of 1.16-1.36 m/s. The code (https://github.com/ylee-sling/Aero-VPA), the trained policy (72.6 MB, https://huggingface.co/ylee-sling/aero-vpa) and the full benchmark dataset are public.
Open source S6
Source：arXiv Robotics — research abstracts · arxiv.org