CBAM meets DropBlock: enhancing robot steering-angle prediction with hybrid attention and structured dropout
Jing Niu, Guanghao Gao, Chuanyan Shen, Jiahao Zheng, Shifeng Liu, Yibo Wang, Jiapei Wei · 2026
Read CBAM meets DropBlock: enhancing robot steering-angle prediction with hybrid attention and structured dropout by Jing Niu, Guanghao Gao, Chuanyan Shen, Jiahao Zheng, Shifeng Liu, Yibo Wang, Jiapei Wei (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13470123.txt.
Description
Jing Niu and colleagues built a lightweight steering predictor in 2026 for robots navigating complex scenes. End-to-end driving systems often miss steering accuracy when clutter, textures, and sky distract from lanes and obstacles. Their three-stage network combines a multi-scale feature pyramid, decoupled channel-spatial attention, and dynamic region dropout.
Channel weights from pooled features strengthen lane and obstacle cues, while gated spatial heat maps mask non-critical regions through DropBlock, suppressing road texture and sky. The cooperative design reports strong steering prediction in tests, suggesting broad use in wheeled robot navigation. Lightweight construction should suit onboard robots with limited compute.
Catalog record
| CREATOR | Jing Niu, Guanghao Gao, Chuanyan Shen, Jiahao Zheng, Shifeng Liu, Yibo Wang, Jiapei Wei |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |