The Worldwide Journal of Robotics Analysis, Forward of Print.
Typical cameras employed in autonomous car (AV) methods help many notion duties however are challenged by low-light or excessive dynamic vary scenes, hostile climate, and quick movement. Novel sensors, corresponding to occasion and thermal cameras, provide capabilities with the potential to handle these eventualities, however they continue to be to be absolutely exploited. This paper introduces the Novel Sensors for Autonomous Automobile Notion (NSAVP) dataset to facilitate future analysis on this subject. The dataset was captured with a platform together with stereo occasion, thermal, monochrome, and RGB cameras in addition to a excessive precision navigation system offering floor fact poses. The information was collected by repeatedly driving two ∼8 km routes and contains diverse lighting situations and opposing viewpoint views. We offer benchmarking experiments on the duty of place recognition to exhibit challenges and alternatives for novel sensors to reinforce crucial AV notion duties. To our information, the NSAVP dataset is the primary to incorporate stereo thermal cameras along with stereo occasion and monochrome cameras. The dataset and supporting software program suite is on the market at https://umautobots.github.io/nsavp.
发布者：Spencer Carmichael，转转请注明出处：https://robotalks.cn/dataset-and-benchmark-novel-sensors-for-autonomous-vehicle-perception-2/
