amap-cvlab/ABot-Recon

English | 中文 [](https://arxiv.org/abs/2608.27529) [](https://github.com/amap-cvlab/ABot-Recon/blob/main/ABot-Recon-Tech-Report.pdf) [](https://amap-cvlab.github.io/ABot-Recon-html) [](https://github.com/amap-cvlab/ABot-Recon) [](https://huggingface.co/acvlab/ABot-Recon) [](https://modelscope.cn/models/amapcvlab/ABot-Recon) [](https://modelscope.cn/studios/amapcvlab/ABot-Recon) [](https://huggingface.co/spaces/acvlab/abot-recon-streaming-3d) [](LICENSE) Long-horizon streaming reconstruction is often approached by adding increasingly elaborate mechanisms for retaining and fusing long-range state. ABot-Recon takes a deliberately local route. At each time step, it solves the same bounded prediction problem: This design keeps model-state memory and per-frame computation independent of the elapsed sequence length. A lightweight motion-visual rotation refiner and composition-aware pose loss are used to limit drift when local poses are composed over long horizons. The full paper reports camera-pose results on KITTI, Oxford Spires, and VBR, together with dense reconstruction on 7Scenes, TUM-Dynamic, and Oxford Spires. The released configuration targets Linux, Python 3.10 or later, PyTorch 2.5.1, and CUDA 12.1. The release environment was validated on NVIDIA A100, while the paper's runtime benchmark uses an NVIDIA H100. ABot-Recon uses paged KV-cache operators from FlashInfer when they are available and falls back to…