门
全向照相机
光学(聚焦)
计算机科学
高级驾驶员辅助系统
无人机
平滑度
实时计算
工程类
全向天线
嵌入式系统
人工智能
电信
生物
天线(收音机)
操作系统
光学
物理
遗传学
数学分析
数学
作者
Christian Scharfenberger,Samarjit Chakraborty,John Zelek,David A. Clausi
标识
DOI:10.1109/itsc.2012.6338635
摘要
A majority of driver assistance systems focuses on assisting the driver when the car is in motion. There is relatively less, to almost no work on assistance systems when the car is stationary. The focus of this paper is on such a system and describes the design and realization of a novel, camera-based anti-trap protection system for smart car doors. Our system uses a single omnidirectional camera that is integrated in each inner door of a vehicle, and detects obstacles such as fingers, hands or legs at the A and B-pillar, door sill and car door without any contact. This is an enhancement to currently available systems that require an active contact between obstacles and sensors in the door area for the detection of trapping. The paper also proposes a scheme for re-calibrating the camera position that may change due to vibrations during the life-time of the car door. Experiments on a fully realized car door prototype show that the proposed schemes prevent trapping by robustly detecting obstacles such as small fingers or hands in critical door regions, and output results in real-time. The latter is important when door users slam the car door. A first study with 20 test users showed that our system strongly supports the door users to prevent trapping.
科研通智能强力驱动
Strongly Powered by AbleSci AI