Machine Vision

Matrix Vision: Smart Eyes for Every Blind Spot

GigE Vision Cameras Shortened the Development of a Safety System for Construction Machines

21.03.2011 -

Everyone has seen them at some time in a television documentation about gold and diamond mines in Australia or South Africa: oversized off-highway dump trucks with a loading capacity of up to 150 tons. A disadvantage of these big vehicles is the limited view of the driver and the high risk of unseen workers around. To guarantee safety, the French manufacturer Arcure offers Blaxtair, a safety system which acquires the danger area of a construction machine in three dimensions and searches for human beings in real-time.

This problem of a limited view does not only concern the southern hemisphere. Here, smaller construction machines like hydraulic excavators, wheel loaders or road roller also have countless blind spots. The vast majority of companies count on acoustic reversing alarm in conjunction with commercial vehicles or construction machines, which is even legally required in some European countries. However, acoustic reversing alarms are not a cure-all. On the one hand, the alarms at close range can be very loud and annoying on construction sites at night. On the other hand, the alarm could be drowned out by normal noise on construction sites or in combination with a music player and headphones. To guarantee improved safety, the responsibility for this should be transferred to the construction machine and thereby the driver. That is what the French Manufacturer Arcure from Paris thought when they created the Blaxtair system. This system acquires the danger area of construction machines in three-dimensions and checks if a human being is located in the danger area.

The task to be solved

The tasks that had to be solved before going to market were anything but easy. First, such a system has to differentiate between a human being and an obstacle like an used oil tank. Second, the detection has to be reliable to avoid false alarms. Otherwise, the driver might switch off the safety system if productivity is compromized.

Because of its experience in sophisticated image processing - especially in human shape recognition - Arcure has chosen the French R&D laboratory CEA (www.cea.fr) as a partner. Arcure and CEA worked together on the principles of associating stereovision techniques with human being recognition algorithms, which successfully led to the Blaxtair system at the end of 2010.

The Blaxtair system

Blaxtair is based on a stereoscopic vision system, containing two digital Gigabit Ethernet cameras as well as a powerful real-time image processing systems. It can detect human shape regardless of position: standing or kneeling. The mentioned algorithms are used to scan the 3D data and to process whether a person is located in a specific area. The computation is done with a pace of 10 pieces of information per second. If an obstacle was detected, the driver will be alerted within 300 ms after the event. The alarm is done both optically (flashing light and colored flat screen) and acoustically (alarm). Three different statuses are available: green for "cleared zone", yellow for "warning event", when a human being is in sight, and red for "danger zone" when a human being is too close.

Suitable camera

However, a suitable camera still had to be found for the final system. Because of the distance between the control unit located in the cabin and the camera system, the flexible and low-priced Gigabit Ethernet was the only interface that came into question. Arcure made long tests of several digital cameras from different manufacturers and finally chose the mvBlueCOUGAR-X from Matrix Vision. Decisive factors in this were several features of the camera that until then were unique in the automation area and still are unique to some extent. Especially the accurate synchronisation of master and slave cameras with the help of timestamps as well as the automatic exposure and gain control (AEC and AGC) of master and slave cameras represented elementary functionality which was needed by the stereo system. Arcure was also convinced by the automatic control of the gain and the exposure due to the result of an area of interest (AOI). Furthermore, placing the mvBlueCOUGAR-X in a closed IP69K housing turned out to be unproblematic due to the low power consumption and the small and robust housing of the camera. Even a temperature of 70°C (158°F) for several hours led neither to any system failure nor to permanent electronic damage. "The fast integration of additional features as well as the easy manipulation and the quite large camera settings choice of the software interface have strongly reduced the electronic research and development phase", says Patrick Mansuy, president of Arcure. "Without this flexibility of Matrix Vision - not only related to the hardware - this would not have been possible", continues Mansuy.

Conclusion

In the future blind spots of construction machines will be a thing of the past. Blaxtair is a high-class, retrofittable safety system and is more powerful and reliable than all other existing systems, also thanks to the GigE Vision camera mvBlueCOUGAR-X from Matrix Vision. It detects and warns about human beings in close range of the unit, but does not bother the driver with unnecessary false alarms. The Blaxtair system allows the driver to increase his working efficiency given that he is informed about persons in the surrounding area of the construction machine in real-time. The accident level can thereby be reduced to a minimum on construction sites.

Contact

Balluff MV GmbH

Talstr. 16
71750 Oppenweiler
Germany

+49 7191 94 32 0
+49 7191 94 32 288

Digital tools or software can ease your life as a photonics professional by either helping you with your system design or during the manufacturing process or when purchasing components. Check out our compilation:

Proceed to our dossier

Digital tools or software can ease your life as a photonics professional by either helping you with your system design or during the manufacturing process or when purchasing components. Check out our compilation:

Proceed to our dossier