Fly360: Omnidirectional Obstacle Avoidance within Drone View
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917319074119680 |
|---|---|
| author | Zhang, Xiangkai Zhang, Dizhe Cao, WenZhuo Wan, Zhaoliang Niu, Yingjie Qi, Lu Yang, Xu Liu, Zhiyong |
| author_facet | Zhang, Xiangkai Zhang, Dizhe Cao, WenZhuo Wan, Zhaoliang Niu, Yingjie Qi, Lu Yang, Xu Liu, Zhiyong |
| contents | Obstacle avoidance in unmanned aerial vehicles (UAVs), as a fundamental capability, has gained increasing attention with the growing focus on spatial intelligence. However, current obstacle-avoidance methods mainly depend on limited field-of-view sensors and are ill-suited for UAV scenarios which require full-spatial awareness when the movement direction differs from the UAV's heading. This limitation motivates us to explore omnidirectional obstacle avoidance for panoramic drones with full-view perception. We first study an under explored problem setting in which a UAV must generate collision-free motion in environments with obstacles from arbitrary directions, and then construct a benchmark that consists of three representative flight tasks. Based on such settings, we propose Fly360, a two-stage perception-decision pipeline with a fixed random-yaw training strategy. At the perception stage, panoramic RGB observations are input and converted into depth maps as a robust intermediate representation. For the policy network, it is lightweight and used to output body-frame velocity commands from depth inputs. Extensive simulation and real-world experiments demonstrate that Fly360 achieves stable omnidirectional obstacle avoidance and outperforms forward-view baselines across all tasks. Our model is available at https://zxkai.github.io/fly360/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_06573 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Fly360: Omnidirectional Obstacle Avoidance within Drone View Zhang, Xiangkai Zhang, Dizhe Cao, WenZhuo Wan, Zhaoliang Niu, Yingjie Qi, Lu Yang, Xu Liu, Zhiyong Robotics Artificial Intelligence Obstacle avoidance in unmanned aerial vehicles (UAVs), as a fundamental capability, has gained increasing attention with the growing focus on spatial intelligence. However, current obstacle-avoidance methods mainly depend on limited field-of-view sensors and are ill-suited for UAV scenarios which require full-spatial awareness when the movement direction differs from the UAV's heading. This limitation motivates us to explore omnidirectional obstacle avoidance for panoramic drones with full-view perception. We first study an under explored problem setting in which a UAV must generate collision-free motion in environments with obstacles from arbitrary directions, and then construct a benchmark that consists of three representative flight tasks. Based on such settings, we propose Fly360, a two-stage perception-decision pipeline with a fixed random-yaw training strategy. At the perception stage, panoramic RGB observations are input and converted into depth maps as a robust intermediate representation. For the policy network, it is lightweight and used to output body-frame velocity commands from depth inputs. Extensive simulation and real-world experiments demonstrate that Fly360 achieves stable omnidirectional obstacle avoidance and outperforms forward-view baselines across all tasks. Our model is available at https://zxkai.github.io/fly360/ |
| title | Fly360: Omnidirectional Obstacle Avoidance within Drone View |
| topic | Robotics Artificial Intelligence |
| url | https://arxiv.org/abs/2603.06573 |