Industry robots can still learn something from the flexibility and responsiveness of football and healthcare robots. So says TU Eindhoven, which will work with industry companies to investigate how their robots can help industry robots gain more insight.
Dealing with unexpected - read: not pre-programmed - events is a matter of course for its football and healthcare robots, according to TU Eindhoven; after all, you never know what your opponent or client is going to do. Industry robots also benefit from greater independence and self-reliance. The university is therefore starting a groundbreaking research project with industry partners to give mobile industry robots an open world view, in which as little is predetermined as possible. Important in this is that the robots learn to understand what they see so they can better cope with surprises.
Many applications
TU/e's High Tech Systems Center will invest 1.5 million euros in the project together with the companies Lely Industries, Vanderlande Industries, ExRobotics (ImProvia), Diversey and Rademaker supported by a contribution from Top Sector HTSM (TKI allowance). It will start this month and last for a total of four years. TU/e will appoint four new PhD students for it, who will be supported by dozens of students. The project has been named FAST, which stands for new Frontiers in Autonomous Systems Technology. The industrial partners who have joined the project are active in all kinds of different applications: agriculture, internal logistics, inspection robots, bakery systems and professional cleaning. In addition, the initiators are open to engaging even more partners.
The project should produce mobile robots that can deal more flexibly with an environment where situations change. This has all kinds of advantages. For example, the environment does not have to be delimited. No specific jobs or landmarks are needed. It also reduces the development time of robots.
Consulting
To realise the 'progress' of robots, the parties will jointly create a 'semantic world model' for robots. Semantic means that a robot can link information to its perceptions, recognise situations and objects, enabling better responses. As an example, being able to distinguish between a pallet of goods, a group of people or a forklift, provides a wide range of possibilities for the robot to respond to. It should also actively seek information itself to better assess a situation, and ultimately perform its task better. For example, by actively surveying the situation with multiple sensors, by extracting information from the Internet of Things, or by consulting a human.
With this new information, the robot completes its worldview. The robots will thus become learning, so that they can get through the same situation more easily next time, or know how to avoid it. They will also increasingly operate between and with humans. Within the semantic world model, interaction with humans is therefore an important part.
Shared issues
The project originated from a survey by the TU/e High Tech Systems Center towards Dutch high-tech industry to find shared issues in the field of robotics where TU/e's expertise can help. In particular, this revealed a desire to make mobile robotic systems more autonomous, which is an area in which the university is leading the way. The teams around the football robots and healthcare robots have both been winning prizes every year for ten years at the RoboCup, the tournament for autonomous robotics.
Competitive position
The cooperation between the university and the industrial partners, and between the partners themselves, allows for the rapid exchange of experiences, and the provision of relevant use scenarios early in the development process. It is expected that this will most certainly contribute to shortening the development time, and thus strengthening the Netherlands as a top player in the field of mobile robotics. The project also provides a platform for students to gain relevant industry experience, which will also contribute to the Netherlands' competitive position given the growing shortage of top technical talent.
At TU/e, the faculties of Electrical Engineering, Mechanical Engineering and Industrial Engineering & Innovation Sciences are participating in the project.
Source: TU Eindhoven
Photo: Bart van Overbeeke
