Driver assistance systems are particularly in demand when the person at the wheel has to concentrate fully on their work - for example when moving bulky loads. This is because the driver often lacks a complete overview of the route and of impending dangers or obstacles in the vicinity. Typical areas of application for driver assistance systems can therefore be found in transport logistics, agriculture, on construction sites, in mines and ports - in other words, wherever industrial trucks, loaders, tippers or container vehicles, some of which are gigantic in size, dominate the scene. But how can the areas that are hidden from human view due to blind spots be safeguarded?

Entire images without time offset

A three-dimensional image is captured either on the basis of stereoscopy using two lenses or using time-of-flight measurement with one lens. In both cases, mechanically moving parts are not necessary. 3D laser scanners, on the other hand, work with rotating mirrors or rely on the movement of objects in front of the scanner. This makes them more susceptible to vibrations and shocks, which are commonplace in harsh industrial environments such as mines or quarries. Static scenes also limit the possible applications of 3D laser scanners due to their design.

The absence of mechanically moving parts therefore increases the reliability and availability of the 3D snapshot sensors. The recording principle is accompanied by a robust camera design tailored to the operating conditions, including a shock- and water-resistant housing. The sensor head of 'Visionary-B', for example, can be cleaned with a fire hose.

Reacts quickly

The following scenario is conceivable, for example: an employee steps out of a bend in a mine tunnel while a mining vehicle is reversing. The sensor system must work correspondingly quickly and inform the driver of the danger immediately. As 'Visionary-B' records several images per second, the basis is created for a high information content of the stereo images, which a computer combines and evaluates to form a three-dimensional image.

This data density is necessary in order to reliably detect hazards or obstacles. However, drivers must not be distracted by this flood of information. The algorithms of the device software therefore filter out all information that is irrelevant to the driver - for example, small bumps and stones as well as rain or fog. In this way, the system only warns the driver of really critical situations.

Scaled solution also for indoors

Visionary-B' captures several images per second and thus creates the basis for a high information content of the stereo images. A computer merges the data into a three-dimensional image and evaluates it.

© Sick

While 'Visionary-B', with its robust design and low lighting requirements, is primarily designed for outdoor use, 'Visionary-T' is available for indoor driver assistance systems. Here, the camera technology also works with intelligent analysis software. However, the distance and size information of objects is not determined by measuring the disparities in the stereo image, but by time-of-flight. 'Visionary-T' has a single-lens design and uses infrared light sources to measure the light travel times between sensed object surfaces and the camera. Here too, the sensor provides depth information for each pixel in real time.

The system measures the fine time differences that the light needs to travel from its source to the object and back again, so that the process can deliver up to 50 three-dimensional images per second. The differences allow the distance to the reflective surface to be calculated. And this data in turn forms the basis for calculating three-dimensional representations in which different colors show different distances.

Thanks to the integrated active illumination of the surroundings, the camera even works in complete darkness. The limits of 3D Time-of-Flight technology are found in areas of application with strong external light sources, such as direct sunlight in open production or logistics areas. This is why the classic stereo camera is ideal for outdoor areas, where neither reflections nor direct sunlight can jeopardize surveillance.

Indoors, however - for example in large logistics centers or extensive production areas - a reliable detection system can be installed with 'Visionary-T' and 3D snapshot technology. The possibilities of the devices go beyond mere collision warning, as they can also be used in collaborative production environments where people work in the direct vicinity of robots. Here, the sensors can be parameterized and used in a targeted manner so that a robot can detect who or what is approaching it. If it is an automated guided vehicle (AGV), for example, the robot continues to move at full speed. However, if the detected object deviates from the taught-in AGV contours, for example because a person is approaching, precisely defined scenarios can be set up - from restricted or slowed movement to a complete standstill. In addition to objects, gestures can also be recognized - an essential basis for human-machine collaboration.

Further application examples for 'Visionary-T' can be found in logistics and material flow: The sensors can be used to reliably determine the dimensions of parcels or packaging units, for example. In addition, the 3D snapshot technology can be used to reliably detect positions when storing and retrieving pallets in high-bay warehouses. During palletizing itself, the 3D sensors reveal free areas. This means that material handling can be better controlled and optimized.

Author:
Fabian Zimmer is Product Manager 3D Compact Systems at Sick in Waldkirch.