A single onboard vision-language-action model flew a quadrotor in four real environments at 48 percent success with no real-world training data
AspireAI community · arXiv
Human cognition draws on a shared perceptual foundation to understand environments, attend to referenced entities, reason about spatial relationships and act, but aerial robot systems are usually specialized per capability, depend on offboard computation, or cannot sustain closed-loop operation on constrained platforms. OpenFlyV2 is a unified onboard vision-language-action architecture using one shared vision-language representation for semantic understanding, language-grounded perception, spatial reasoning and continuous aerial control. A compact action expert converts the shared multimodal context into short-horizon flight waypoints, while motion-aware temporal memory and action priors reduce the cost of closed-loop execution. The authors construct a simulation-based dataset of approximately 70,000 navigation trajectories and 60,000 vision-language instruction-response pairs. OpenFlyV2 generates action chunks of 12 waypoints entirely onboard at 10 Hz on an NVIDIA Jetson Orin NX, and transfers zero-shot from simulation to physical quadrotor flight, achieving a 48% navigation success rate across four real-world environments without real-world aerial training data. The limitation is that 48% success is well short of reliable autonomous flight, so this demonstrates transfer rather than a deployable navigation stack.
Read the story and explore more from the AspireAI community.
Two sparse-autoencoder analyses have resisted scaling: matching semantically similar features across layers, and compressing large feature circuits into interpretable supernodes. This paper argues both are the same underlying problem, estimating semantic distances between SAE features that sit on different activation manifolds. Its representation treats each feature not as a single decoder vector but as an activation-weighted distribution over the hidden states expressing it, projects those distributions into a shared reference space, and compares them with Wasserstein distance. The representation is proved invariant to activation rescaling and stable under perturbation, and recovers true matches under finite-sample margin conditions. Empirically it outperformed decoder-vector and LLM-based baselines and captured subtle functional distinctions between related features, and it compressed large feature circuits into interpretable supernodes automatically. The limitation is that this is interpretability tooling: the paper demonstrates better feature matching and circuit compression, not that downstream interpretability claims become more accurate.