A Customized Vision System for Tracking Humans Wearing Reflective Safety Clothing from Industrial Vehicles and Machinery

Abstract

This article presents a novel approach for vision-based detection and tracking of humans wearing high-visibility clothing with retro-reflective markers. Addressing industrial applications where heavy vehicles operate in the vicinity of humans, we deploy a customized stereo camera setup with active illumination that allows for efficient detection of the reflective patterns created by the worker's safety garments. After segmenting reflective objects from the image background, the interest regions are described with local image feature descriptors and classified in order to discriminate safety garments from other reflective objects in the scene. In a final step, the trajectories of the detected humans are estimated in 3D space relative to the camera. We evaluate our tracking system in two industrial real-world work environments on several challenging video sequences. The experimental results indicate accurate tracking performance and good robustness towards partial occlusions, body pose variation, and a wide range of different illumination conditions.

IR tag detection and tracking with omnidirectional camera using track-before-detect particle filter

“…[9] It uses shape information to detect the clothing, so the detection performance can highly depend on the pose and relative position of the target.…”

The Next Step in Robot Commissioning: Autonomous Picking and Palletizing

“…The successful integration between the described sub-systems then depends on the application specifics and target requirements. Of these components, only the manipulation framework has not undergone prior evaluation (for tests of the navigation system see [21], [5], [3], for the AGV safety features see [6]). The manipulation system was evaluated with a stationary base in a two-step manner: first, we characterized the robustness of our grasp control scheme in isolation.…”

“…As most of our approaches to AGV navigation [5], [3] and human detection [6] have been published previously, here we only give a review and describe relevant extensions while focusing mainly on the manipulation aspect. The key insight is that the relatively simple structure of autonomous pick & place tasks allows fast and robust grasping by coupling compliant manipulator control strategies with a novel redundant grasp representation.…”

“…For localization, a Velodyne HDL-32E 3 Lidar (4) is employed. On-board safety systems include a back-mounted SICK S300 4 laser scanner, a dedicated camera system (6) to track human workforce wearing reflective safety clothing and a projector which visually indicates the AGV's intended path (5). A lightweight KUKA LBR iiwa 5 manipulator (3), fitted with an under-actuated gripper (2) with conveyor belts on the inside of each finger, is used for robust grasping and object manipulation.…”

Sistema automatizado para monitorear el uso de equipos de protección personal en la industria de la construcción

“…En laúltima década, la detección de EPP ha sido utilizada como filtro inicial en procesos de seguimiento visual de trabajadores en sitios de construcción. Este filtro es un clasificador de personas trabajadoras basado en la hipótesis de que todo trabajador objetivo debe utilizar ropa de alta visibilidad (Park and Brilakis, 2016;Seong et al, 2017;Mosberger et al, 2014;Konstantinou et al, 2019). La clasificación de estos EPP se materializa mediante detectores de color, que pueden diseñarse en distintos modelos como RGB, HSV o Lab, para determinar en una imagen la existencia de objetos con colores de seguridad conocidos (Park and Brilakis, 2016;Seong et al, 2017;Konstantinou et al, 2019;Mosberger et al, 2014).…”