Fully AI-Controlled Vehicles: How Close Are We?

How can artificial intelligence (AI) and automation revolutionise the way heavy mobile machinery operates? What challenges remain in making autonomous vehicles truly safe and efficient? To explore these topics, we spoke with Prof. Reza Ghabcheloo, an expert in autonomous mobile machines at Tampere University (Finland), whose research focuses on advancing intelligent and automated systems for industrial applications.

Q: Could you briefly introduce the Autonomous Mobile Machines research group you are working in at Tampere University? When and why was it established, and how has it evolved over time?

I completed my PhD in marine robotics in Lisbon, but when I moved to Tampere in 2008 for a postdoctoral position at the Centre of Excellence in Generic Intelligent Machines, I quickly realised that Tampere was a global hub for mobile working machines. These are heavy-duty machines used in industries such as construction, mining, forestry, agriculture, and port operations – essential for economic and industrial progress. In fact, some of the most advanced intelligent mobile working machines originate from Tampere.

As a classical control engineer, I initially focused on ensuring these machines could follow precise routes and reach targets accurately, even in challenging conditions with sensor noise or camera failures – for example, automating a warehouse forklift to fetch pallets. Over time, our research expanded to more complex tasks, such as autonomous earthmoving and navigating uneven terrain, which required high-precision decision-making and advanced AI techniques. Today, our work integrates sophisticated modelling, high-dimensional sensor inputs like LiDAR and cameras, and real-time decision-making to enhance machine autonomy.

However, one of the biggest challenges remains safety. AI-based control systems for critical applications – such as autonomous vehicles and industrial machinery – are prone to errors that humans would not make. Just today, in January, I was writing back to a friend, typing “I hope you had a …” and AI suggested “a great summer.” Imagine if AI made such suggestions for a car driving on a road.

This unpredictability makes AI behaviour difficult to explain and verify using traditional methods. Thus, we are actively working to address these safety concerns.

Q: Speaking of AI in autonomous vehicles: it has been a trending topic in both science and public discourse. Could you share insights into the current state of the field and future trends?

The promise of autonomous vehicles is clear: reducing road congestion, lowering accident rates, and improving logistics efficiency. Many companies have invested heavily in this technology over the past decade. However, they have also realised that achieving full autonomy is much more challenging than initially anticipated. As a result, we have seen an increase in partnerships among major industry players, while many smaller startups have struggled or gone bankrupt.

That said, driver assistance technologies have already become standard. Features such as emergency braking, lane-keeping assistance, and automatic high beams are now included even in base-model vehicles. While these systems enhance convenience, they do not yet function as full-fledged safety features because they do not work in all conditions. Achieving reliable full autonomy is still a long journey, but progress is being made.

Predicting where we will be in 10 years is difficult. However, advancements in onboard computing power, sensor technologies, AI infrastructure, and connectivity are accelerating innovation. I anticipate that within the next decade, we will see automation becoming commonplace in controlled industrial environments, such as warehouses and ports, as well as in low-speed, low-risk transport applications. However, economic factors and global political developments could significantly influence the pace of adoption.

Q: In your opinion, what role do universities, scientists, and projects like SustAInLivWork play in the development and adoption of AI in industry and society?

Universities and independent research projects like SustAInLivWork play a crucial role in bridging the gap between AI research and real-world applications. AI has already demonstrated its potential in industries such as transportation, energy, and healthcare, but its full capabilities remain underutilised.

As AI is evolving rapidly, uncertainty remains about what is truly achievable. Academic institutions serve as neutral spaces where industry leaders can gain objective insights and make informed decisions about adopting AI-driven solutions. By fostering interdisciplinary collaboration and knowledge exchange, we can help ensure that AI is integrated into society in a way that maximises its benefits while addressing safety and ethical concerns.