The toolbox of robotics

Especially for Germany, the growing demand for robotics solutions across all sectors is a great opportunity. The local industry, which is characterized by a deep technical understanding and a pronounced depth of production, seems to be made for a high technology such as Advanced AI Robotics.

At the Tech Solution Provider IAV, too, numerous engineering teams are dedicated to researching and developing the required technologies. The harvesting robot is only one of several applications.

First walk tests in robotics

Paul, the stair climber, is a first-time project in the field of autonomous robotics systems from IAV. About eight years ago, the engineering team developed Paul as a friendly assistant for the shopping. Paul drives on two wheels, finds space in the trunk and can climb stairs along with the shopping bags. Since then, IAV has gone deeper into the subject of robotics, with the aim of developing scalable and flexible technology solutions in hardware and software.

IAV Roboter 2311
The stair climber Paul has been one of the first robotics projects at IAV.

Two important milestones along this path are the MiRo Base and EnerGlider research projects implemented together with universities and industrial partners under the project management of Dr. Manus Thiel. MiRo Base is a system platform for modular, remote-controlled underwater vehicles, while the EnerGlider, an innovative high-altitude wind turbine, is a low-maintenance and efficient alternative to conventional wind turbines. “To this day, we have been able to draw on the findings of our MiRo Base project for guidance, navigation & control (GNC) topics, with which we have successfully completed diving trips in the Baltic Sea. The test flights with the EnerGlider have also brought us forward in this respect,” says Dr. Thimo Oehlschlägel, team manager Control Engineering Excellence Cluster at IAV.

In the field of image recognition and environmental perception (Computer Vision), the “Franka” concept made decisive progress. This is an arm that can grab and hand over tools autonomously. “In order to perform these actions, Franka must be able to recognize individual tools, differentiate them from each other using AI, and calculate his own position and distance between himself and the tool. Then the arm reaches up and swings toward man. The results of this initial project can also be found in our current harvesting robot in a more developed form,” says Professor Dirk J. Lehmann, specialist for Computer Vision and visual data analysis at IAV.

Roboterarm 12547
A robotic arm in action (symbolic picture).

Modular robotics kit as a platform for new innovations

The goal of the development teams in the field of robotics is to develop basic technologies and basic methods that can be transferred as best possible modules into future projects and optimized there. Only in this way can developments be scaled and the numerous fields of application for robotics solutions, such as agriculture, maintenance or automated monitoring, can be filled with agility. That’s why the experts are also working on a software framework for robotics that is constantly being further developed. “We can quickly integrate functions such as object recognition, distance evaluation and image processing into other projects on the basis of this framework,” says Lehmann.  He and his team see the greatest challenges in the coming years in the areas of onboard computing, data fusion and AI online training. “Those who meet the basic requirements in these areas are able to gain faster access to high-level functions for highly automated systems, especially with regard to AI-driven decision-making processes. With the right tools in our toolbox, we are able to respond to customer requests in the shortest possible time.” This proves the short development time of just three and a half months for the first prototype of the harvesting robot.

The modular robotic kit approach also works well when you take a look at the beekeeper robot “TomBigBee”, which has some similarities to the harvesting robot. “The common feature of both systems is that they drive automatically on a rover. In both cases and with numerous other mobile highly automated systems, our state-of-the-art navigation algorithms, which have already been tested in flight applications, as well as lane and path planning algorithms, can be used with little porting effort. In addition, our beekeeper robot also has image recognition to locate the queen of bees and check the beehives for possible mites,” summarizes Oehlschlägel. The best conditions for continuing the journey, which officially began eight years ago but has only just begun.