Multi-level on-board data fusion for 2D safety enhanced by 3D perception for AGVs
| Autor: | Sergiu Nedevschi, Christian Stimming, Annette Krengel, Markus Boehning, Andrei Vatavu, Szilard Mandici |
|---|---|
| Rok vydání: | 2015 |
| Předmět: |
Truck
Computer science business.industry ComputerApplications_COMPUTERSINOTHERSYSTEMS computer.software_genre Sensor fusion Obstacle Synchronization (computer science) Computer vision Artificial intelligence Omnidirectional antenna business computer Collision avoidance Stereo camera Data integration |
| Zdroj: | ICCP |
| Popis: | Modern AGVs are equipped with several safety laser scanners with a combined 360 deg field of view around the AGV to detect and subsequently avoid collisions with other AGVs, structural elements and, most importantly, workers. This contactless environment perception approach fulfils current safety legislation and safety regulations for driverless industrial trucks. However, obstacle detection is limited to a 2D plane parallel and close to the ground, unable to detect protruding or hanging objects in the path of the AGV. In order to avoid collisions with these kinds of objects as well, the idea of PAN-Robots is to enhance the existing 2D safety by a 3D perception system based on an omnidirectional stereo camera. This paper describes the multi-level on-board sensor data fusion strategies implemented in the PAN-Robots project. The fused information of tracked and classified objects is not only used for on-board risk assessment and emergency collision avoidance, but is also communicated to the global control center for advanced fleet coordination and intelligent AGV navigation. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|