European project aims to impart robots with human-like vision

Citation

(2010), "European project aims to impart robots with human-like vision", Industrial Robot, Vol. 37 No. 5. https://doi.org/10.1108/ir.2010.04937eab.002

European project aims to impart robots with human-like vision

Article Type: News From: Industrial Robot: An International Journal, Volume 37, Issue 5

February 2010 saw the launch of an ambitious project which aims to improve dramatically the visual capabilities of robots to enhance capabilities such as autonomous navigation. Dubbed three dimensional adaptive camera with object (TACO detection and foveation), this 36-month, collaborative project is part of the EU’s 7th Framework Programme and involves four research institutes, two industrial companies and one university from four European countries.

The broad aim is to allow robots to play a major role in fields that are currently dominated by humans, such as cleaning, construction, maintenance, security, healthcare, entertainment and personal assistance. However, to achieve this, robots will need to understand and make better sense of the environment in which they are navigating. To date, the approach has been to use video cameras and shape-and pattern-recognition software but the capabilities of this are severely limited, partly because they do not allow the robot’s “brain” to become an active part of the process. TACO seeks to impart robots with a more human-like and efficient type of vision through the use of a technique termed 3D foveation. In human vision, the resolution is highest at the point of gaze and falls rapidly towards the periphery and this inherent multi-resolution perception is an example of foveation. In artificial systems such as robots, this involves the acquisition of coarse 3D images of what is in the robot’s field of view and then applying a fast image-recognition technique to this preliminary image. As this is done, specialised software identifies areas of interest in the image and then focuses the full power of the robot’s visual capabilities on those targets. The rest of the image, which is deemed to contain irrelevant information, is then ignored. This will be achieved through the use of micro-mirror MEMS technology combined with a time-of-flight sensor in a system that will be small, light and easily mounted on an ordinary sized service robot or a robotic arm. An advantage to using foveated sensor hardware is that the image collection and encoding can occur much faster than in a system that post-processes a high-resolution image in software. According to Jens Thielemann, the Project’s Technical Leader, the TACO sensor will enable significantly better, faster and cheaper 3D sensing compared to current laser scanners. Through the foveation process, the sensor will provide ten-times better resolution than existing devices.

The three specific scientific and technological objectives of the project are stated as:

1. 1.

   to develop a flexible, compact, robust and low cost 3D imaging device providing high-resolution 3D data of high quality;

2. 2.

   to achieve 3D measurements of increased spatial and temporal resolution in detected regions of interest by developing adaptive and intelligent software for sensor control and 3D foveation; and

3. 3.

   to benchmark the 3D sensing system on robots in an everyday environment test bed with interaction with everyday objects.

The final prototype for validation and testing will be completed in month 30 and following this phase, the project aims to offer the new technology to the European robotics industry and make it attractive to both large and small companies. Further, the consortium plans to make TACO knowledge highly visible within the industry and the scientific community.

For more information, see the TACO project web site: www.taco-project.eu/ or contact the Project Co-ordinator: Klaus-Michael Koch, e-mail: koch@technikon.com

Technikon Forschungs-und Planungsgesellschaft, Austria – Coordinator.

The Shadow Robot Company Ltd, UK – Robot gripper applications.

Oxford Technologies Ltd, UK – Applications in harsh environments.

Technische Universitaet Wien, Austria – benchmarking with state-of-the-art 3D sensing methods.

Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung E.V., Germany – Micromirror scanning device, 3D range sensors.

Stiftelsen SINTEF, Norway – 3D foveation software.