A smarter, safer and bias-free digital world in the future

Artificial intelligence (AI), robotics and the Internet of Things (IoT) are transforming our world at a quick pace. Yet, there is still some way to go before we can take advantage of the full potential of these technologies. Complex challenges related to the performance and trustworthiness of digital systems still need to be resolved.

With their sights set on a more trusted and connected world, the EU-funded projects ALOHA, ENACT and PDP4E are tackling the issue of secure and smart IoT systems from different angles. ALOHA is focusing on deep learning, ENACT on smarter IoT, and PDP4E on better privacy and personal data protection.

“To realise the massive promise of an IoT-driven world, we must solve complex challenges,” says Otto Berkes, Chief Technology Officer of CA Technologies, a partner in all three projects, in an interview in ‘Robotics Business Review’. “These hurdles must be overcome before we can deliver IoT systems that can provide valuable and trusted data, be adaptable and open to new technologies – systems that haven’t even been invented yet.”

Towards bias-free and ethical AI

There are numerous examples of unintended bias in AI models – from racist Twitter bots to unwelcome Google results – caused by the data the algorithms are based on. Deep-learning algorithms are used in AI to create models that can analyse data in a way that is similar to human logic. With the aid of such algorithms, ALOHA aims to investigate how IoT applications can learn from experience and draw conclusions much like a human would do. The research conducted will focus on creating a more bias-resistant and ethical AI.

Towards smarter connected systems

When IoT systems, devices and applications operate in potentially hostile environments where exposure to cyberattacks or harsh weather is a possibility, securing and maintaining them isn’t a simple matter.

ENACT is developing innovative ideas for new tools needed to create and monitor resilient, smart IoT systems that can be trusted in scenarios that weren’t anticipated during the software development process. “For instance, two devices on a smart train may be sharing critical information on an unencrypted channel,” Berkes writes on the CA Technologies website. “Early detection of this vulnerability may avoid software development delays due to efforts to secure the system.”

The project will also explore AI-based self-diagnosis for smart operation, which will make it possible for trains to detect anomalies and predict and avoid critical situations. Besides intelligent transport systems, ENACT will test IoT systems in two other spheres: e-health and smart buildings.

Toward better privacy and data protection

The recently enforced General Data Protection Regulation (GDPR) was designed to protect the privacy of all EU citizens and reshape the way organisations throughout Europe approach data privacy. PDP4E will develop methods and techniques to help software engineers develop code that complies with the GDPR and other regulations.

All three projects were launched in the first half of 2018. ALOHA (software framework for runtime-Adaptive and secure deep Learning On Heterogeneous Architectures) and ENACT (Trustworthy and Smart Actuation in IoT systems) are expected to be completed by the end of 2020. PDP4E (Methods and tools for GDPR compliance through Privacy and Data Protection Engineering) will end in January 2021.

For more information, please see:
ALOHA project website
ENACT project website
PDP4E CORDIS project web page