Self-driving cars capture everyone’s imagination. Less common knowledge is that the development of the technology to make machinery and other vehicles autonomous is supported in Flanders. Flanders Make, the research center for the manufacturing industry, maintains close partnerships with the industry to study a wide range of applications: from logistic robots, tractors and weaving machines to assembly cells within production areas. The key to success is a simultaneous design combining mechanics, electricity, electronics, software and control. We give the floor to two researchers.
“In future, agricultural vehicles such as harvesting machines will operate fully autonomously,” says Jan Goos, an engineer with Flanders Make. “They will drive autonomously across the field and, at the same time, perform an operational task such as harvesting. Good software is fundamental for the correct operation of these vehicles as it must respond to all changes indicated by sensor signals, be it in terms of crops, soil or weather conditions.”
“Systems are getting increasingly complex because they have to fulfill an increasing number of requirements,” confirms Suzanne Van Poppel, Core Lab Manager with Flanders Make. “Just driving, for instance, no longer suffices. Accordingly, the design, specifications and control technology of these systems are becoming increasingly complex as well. On top of this, the time to market must be reduced. Software helps vehicle or machine developers to make the right choices to reach these objectives at an earlier stage.”
The impact of design changes
A major challenge in the design process is that hardware and software cannot be considered separately from one another. Yet, it’s still standard practice that in a product design process different teams are working separately, all making decisions within their own area of expertise. The hardware engineer selects the components, the control engineer develops the controller and the computer scientist implements the software. To optimize the design process, we need a multidisciplinary approach. Co-design is the key word here.
Suzanne: “At Flanders Make, we focus on the co-design of hardware, software and control technology. We go for an integrated design process that maps the impact of design changes throughout the entire lifecycle of a product. The hardware, for instance, imposes restrictions on the software. It would be a huge waste of time to find out at the end of a design process that the software isn’t working because the hardware isn’t compatible. The same applies to the production phase. How can we adapt the assembly line so as to be able to assemble a wide variety of products and deal with future design changes?”
Such an approach isn’t exactly easier because you have to take into account many more aspects from the very start. When done well, however, it does significantly reduce the number of design iterations. In other words, co-design allows to reduce the time to market of systems and to manufacture customized products as it enables to market highly advanced products in many variants and at the cost of mass production.
Co-designing a drive and controller
Vehicle and machine builders such as Dana Belgium and Michel Vandewiele must be able to deliver customized systems that can perform increasingly well under very diverse loads. Current design methods are sequential: first drive, then controller. This doesn’t always lead to the best performances and to the lowest possible cost.
Jan: “We’ve assessed the fuel consumption of different hybrid powertrains on a virtual track. For this, we had to simultaneously optimize the control system. The result is a more economical powertrain. Another research topic is the design of product variants. Here as well, we use co-design methodologies and tools to realize cost savings.”
“It’s important to adapt the control system to the functional requirements of the drive,” Suzanne adds. “A passenger car must be able to handle very different situations and must meet very different requirements than a vehicle used for mining purposes. The problem with a sequential design is that the functional requirements in the next stage of the design process aren’t always known. When a good design is led to the next stage and it then appears that there are still insurmountable problems, you’re back to square one. This results in delays and considerable design costs or in a system with inconsistencies. We’ve noticed from our talks with companies that they are struggling with this problem and that having the right co-design methodologies and tools would allow them to take a major step forward.”