There is a quiet yet significant revolution underway within the massive electronics industry. Harnessing machine learning (ML) and artificial intelligence (AI), companies within the sector are building new software that saves designers, engineers, distributors and manufacturers time and resources, gradually cutting back tired and analog working methods that were previously used for creating electronic products.
ML and AI are more advanced than ever. But, despite great strides, it is surprising that a technically-established vertical such as electronic engineering is not yet dominating the charge toward automation. For example, printed circuit boards (PCBs), crucial components in all electronic devices, are often still being designed using human engineers’ experiential knowledge and thought processes. Design and manufacturing times for PCBs remain archaically reliant on humans.
But winds of change are sweeping through the industry; ML is beginning to refine design processes. From improving searches for parts and components, to digitizing legacy engineering documents, to assisting in design generation, ML illuminates insights about processes that would otherwise be invisible to engineers.
So what platforms are available to engineers to reduce PCB design process times, and what are their drawbacks and merits?
Let’s start with traditional electrical computer-aided design (ECAD) tools. These are complex software tools designed to allow engineers to perform any kind of detailed design (offering some automation). However, they are usually only tailored to manual engineering work. Examples include Altium Designer, Siemens EDA, Cadence OrCAD, AutoDesk Eagle and Zuken ECAD tools.
An alternative form of assistance that is frequently used, yet is largely inefficient, is the office (or project) tool. Even today, engineers are using office tools such as Excel, Atlassian, Visio and others to manage much of their activities, such as maintaining wikis and managing projects. As they were never designed for day-to-day engineering work, these tools have multiple shortcomings, lacking the specificity necessary to save engineers time when completing electronic designs.
Up-to-date information critical
Database providers additionally offer software tools that give engineers insights into component prices, availability and (some) technical specifications.
In the electronics industry, up-to-date information about components and semiconductors is crucial. However, this information can undercut and even negate engineers’ progress when they are designing products because databases lack details about circuits and reference designs that are absolutely necessary to make composition blueprints into a manufacturable reality.
These previous three examples are all constituent platforms often used by engineers that, individually and collectively, fail to deliver on informational and organizational coherency or time efficiency.
Therefore, there is a distinct necessity for automating platforms, a new class of which have recently entered the market. Cloud-based platforms, focusing on high levels of abstraction and functional design views, provide as much automation as possible and leverage the sharing and collaboration of different engineers. These platforms usually integrate smoothly with existing design tools, such as traditional ECAD.
The power and dangers of data and machine learning’s significance
A ubiquitous topic of the digital age, not simply in electronic engineering, concerns the evolution of ML and AI amid abundant data flows. Technological capabilities for data storage, compilation and comparison have vastly expanded in recent years, and have thankfully shrunk the time and resources that engineers spend on projects. Despite this, data handling remains a difficult proposition as developers receive more and more information.
Without careful management and proper “hygiene” processes in place, more data can mean more issues for those grappling with it. New challenges arise from sheer amounts of data, and particularly bad data. For engineers, having access to billions of datasets is useful up until the point where there are information overloads, which was all too common when PCBs were designed manually, for example.
Data must be channeled in ways that ML is rendered appropriate for use in electronic engineering. The future of the industry, and tech more widely, demands a focus on data quality. Data must be pointedly compacted to make it easily accessible and digestible. Users need clarity on which data points are essential and what they need to do with them. It will fall to data analysts to decipher the masses of data, with these roles then increasingly attracting higher investment from companies in the near future and beyond.
More flexibility, creativity
Within electronic engineering, introducing new data types also fosters more flexibility and creativity. Not only can selecting components and creating functional designs be achieved more quickly, but other design characteristics (such as sustainability) can be interwoven into final schematics.
In sustainable designs, components are selected based on performance, recyclability and longevity, leading to more appropriate sourcing with new data streams becoming more prominent at the design stage.
Ushered in by ML, the overall significance of healthier data management capabilities is the reduction of learning curves required for the industry’s workforce and the corollary effects of this. Ground-level tasks in PCB design previously undertaken by more proficient engineers are now being shifted to less experienced engineers using ML tools. This allows highly trained designers to focus on more specialized tasks and can aid firms with workforce shortages, with ML picking up the slack.
Automation vs. human input
The premium opportunity for AI and ML in electronic engineering is error removal from design and manufacturing processes. Leveraging proven settings and designs from millions of users helps to avoid mistakes and improves versatility. Users can replace components and adjust designs quickly to market conditions and disruptions. AI and ML-informed automation is — and will continue to be — revolutionary for the sector in design time efficiency.
Yet despite the whirlwind advance of automating technology, human input remains paramount. Questions over deploying this technology mustn’t concern what we can automate, but what we should automate. Creativity and innovation in design are not spearheaded by AI but by skilled engineers. If we want to drive innovation in electronics, we will always need the human brain.
What should be automated are the manual and tedious tasks that waste engineers’ time (which could otherwise be spent on more important areas). Full automation is not the final desired state, but it is the turbocharger firing new efficiencies in electronic engineering.
Alexander Pohl is cofounder and CTO of CELUS.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!