Digital minds think alike at Brussels tech conference

The iMinds conference in Brussels explores the latest thinking on digital innovation, while showcasing projects by Flemish researchers that are putting these new ideas into practice.

Danny Goderis and Luc Van den hove, the chief executives of iMinds and imec respectively, will open the conference, setting the scene and talking about the forthcoming merger of the two research organisations. They will be followed by Jamie Bartlett from British think tank Demos, who will talk about the digital underworld lurking out of sight of Google’s search engines.

Subsequent sessions highlight trends that are shaping digital innovation. First comes the data-driven economy, in which new ways of dealing with information create innovative products and services or reshape traditional industries. Then there is the internet of things, the idea that objects can share information over digital networks, transforming the way we live and work.

Break-out sessions cover digital manufacturing, trust in the digital world and how bright ideas become international companies.

Artificial intelligence

Speakers from academia and companies such as IBM and Proximus will address the broad sweep of these issues, but to see what they might mean in practice, you need to check out the booths where iMinds is showing off its research projects.

Several trends come together in the Wonder project, which is looking at new ways of using the Zora health-care robots developed by Flemish company QBMT. The robots are currently used for pre-programmed tasks; the idea is to see if they can work with greater independence.

One approach is to use ideas from the internet of things to extend a robot’s senses. For example, if a robot responds to a call in another part of a nursing home and needs to plot a route, a wireless connection to sensors all over the building could provide the necessary information.

“A robot can’t look around the corner, but if there’s a sensor that indicates a door ahead is closed then it knows and can plan its movements more intelligently,” says Pieter Simoens of Ghent University.

An alternative would be to use the same wireless connection to allow the robot to unlock the door, or to pass the task on to another robot in a better position.

Data about staff movements that could also be gathered in a connected nursing home might also help. “If a robot needs to take an elevator to another floor but can’t activate the elevator itself, maybe we can predict when people will be at the elevator so that the robot can hop on,” says Simoens. Pilot projects in two Ghent nursing homes are currently being set up, with sensor company Xetal.

Ship-shape

An example from the data-driven economy involves the movement of shipping containers along inland waterways. BCTN, which operates inland terminals in the Netherlands and at Meerhout in Antwerp province, approached researchers at the University of Leuven to see if they could make the scheduling of barges more cost-effective.

Currently, barges operate on a semi-regular schedule between inland and port terminals, delivering and collecting containers within set time windows. The researchers were given several months’ worth of data on these movements and set to work designing a computer model to produce better schedules.

This meant dealing with several “real world” complications, such as not knowing exactly how long a journey might take, given factors such as locks on the waterways. Similarly, a container barge cannot be treated like a single vehicle. They often consist of several barges or bakken pushed by a boat. Some may be left behind if not required on a particular leg of the trip, but must be collected again later.

“The bakken make it more interesting,” says Greet Vanden Berghe, who leads the project from the university’s Ghent campus. “They give an additional level of potential for optimisation and academically exciting complexity.”

Finally, an option was added of moving a container by road if that was more cost-effective than the barge.

The model has been tested against academic benchmarks and performed very well. “So we can be more or less certain that we outperform the manual schedulers, who have much less information and no computing power,” says Vanden Berghe. But the real test will be when the company uses the model in practice.

28 April, Square, Mont des Arts, Brussels. Photo courtesy iMinds