Did you know that many incidents happen every year between heavy machines and people walking around? Heavy machines, like those used to move containers in ports, big excavators, etc. offer very poor visibility to the human operator. This is why implementing a People Detector for Heavy Machines is crucial. Whenever a person is located close to the vehicles, his safety is at serious risk.
The purpose of the People Detection System for Heavy Machines was to create a technological demonstrator. It is capable of automatic people detection around reach stackers in harbour areas and construction sites.
The project was partially funded by a leader company in logistic operations, and ran from 2018 to May 2019. A People Detector for Heavy Machines can significantly reduce these incidents.
Even though RFID-based smart bracelets are a very cost-effective, practical solution, many people simply don’t want to wear them. Even when obliged, they “forget” it into their own locker. Thus, a People Detector for Heavy Machines offers a passive safety solution.
Following a series of preliminary training activities that our partner required to be updated to the state-of-the-art of technology for people localization, it asked us to develop a custom embedded system. This system is used to deploy as a supplemental artificial eye, capable of identifying people by using the most advanced artificial vision techniques. This People Detector for Heavy Machines includes system features such as human recognition and distance estimation. It offers different alerts (audio, visual) to warn the driver in different situations.
The fundamental hardware module of the system consists of a MOTEC camera for off-road vehicle applications. Coupled with an NVIDIA TX1 platform, it is mounted on the reference vehicle. The system is camera agnostic, i.e. it is working even with standard analog cameras with a large fisheye effect mounted on offroad vehicles. It requires no special adjustment.
The analog video provided by the camera is acquired by the frame grabber integrated into the NVIDIA TK1 board. Every single frame is processed by the vision-based algorithms developed by Aitronik, to recognize people and locate them in the scene.
The image is then overlaid with a bounding box that is superimposed over the detected persons and sent to the display.
The People Detection System is currently used to help the operator onboard the vehicles detect people and avoid accidents.
Do you have any question? Get in touch with us!
