Autonomous CBRNE mapping and manipulation with exploratory robots

When tasks are too dangerous for humans, robots provide support. CBRNE reconnaissance is a prime example of such a task. CBRNE stands for »Chemical, Biological, Radiological, Nuclear and Explosive«. Examples of such materials are aggressive hazardous materials and radioactive contamination, which pose extreme challenges for the armed forces and police security forces in an emergency. A danger to life and limb can never be completely ruled out, even with the right equipment and extensive training. Robots equipped with CBRNE sensor technology and autonomous assistance functions take over in such cases.

Supported by the German Federal Armed Forces, the »Cognitive Mobile Systems« department is researching the innovative combination of CBRNE reconnaissance sensor technology, navigation strategies and geodata processing. It is also focusing on development of intelligent assistance functions that greatly simplify the remote control of robots by making it more intuitive and easier to use, or even by performing fully automated subtasks.

To ensure precise detection, localization, and mapping of hazardous materials, the scientists link and coordinate sensor data provided by CBRNE detection robots with navigation strategies and additional information from digital geo-databases. For areas that cannot be explored by the robot, a geostatistical estimation method supplements the measured value map in the inaccessible areas.

Potential hazards can thus be more accurately detected, localized and – particularly important for emergency services – precisely mapped. For this purpose, the different data are merged into a common situation map accessible to the control centre, providing it with ongoing decision support.

Many application scenarios still require remote, mobile manipulation of objects, notably when securing hazardous materials. However, remote control of a robot with a gripper arm used to require lengthy operator training. This is changing with the availability of ever more powerful mobile computers and innovative AI algorithms, which enable more intuitive control and, in some cases, even the automated completion of tasks.

With these new technologies, for instance, real-time processing of high-resolution video data streams, so-called »robot vision«, is steadily improving. The importance of these technologies in autonomous high-level applications is rapidly growing. They are also the basis for a further innovative assistance function from the FKIE laboratory: a »click & grasp« system that enables a gripper arm to pick up objects automatically in a live video image with just a click of the mouse and, with another mouse click, place them in another location.

The functionality and efficiency of the mapping and intelligent assistance methods developed by the Fraunhofer FKIE were not only scientifically proven under laboratory conditions, but also successfully displayed with a demonstrator at the newly established »European Robotics Hackathon« (EnRicH).

The kick-off competition, organized by the Fraunhofer FKIE in cooperation with the Armaments and Defence Technology Agency (ARWT) of the Austrian Armed Forces, took place in June 2017 at the decommissioned Zwentendorf nuclear power plant. Its goal was to evaluate and advance the current capabilities of mobile robots to respond to nuclear disasters.

Eleven international teams and their robot systems took part in the challenge posed with a real incident scenario. Only seven of them, among them the Fraunhofer FKIE team, competed in all three categories: »3D mapping«, »radiation mapping« and »manipulation«.

The Fraunhofer FKIE was the only participating team that succeeded in letting its robot autonomously reach for the leaked material, check it for radiation, transport it away and deposit it in a special container. The team was awarded the »Best Manipulation« prize for its achievement in identifying and recovering four out of five radiation sources with this method. The scientists won another first prize in the category »Best Radiation Mapping«.

The research project will be continued in different directions. For example, many additional functions will be implemented based on the comprehensive approach to image evaluation for universally deployable robots, which was already used for the described »click & grasp« system. In this context, powerful sensors and miniaturized computers along with virtual reality glasses and control units with motion detection will come in handy. They are a driving force for robot vision. In the future, not only military technologies but also completely new operating concepts and fields of application will be developed.