From Capability to System: The Next Phase of Off-Highway Autonomy

Over the last few weeks, I’ve been exploring autonomy and electrification across the Off-Highway sector - and one thing has become increasingly clear:

Mining is ahead.

But more importantly the nature of the challenge is changing.

That is the theme I want to explore in this month’s Autonomy Roundup.

Changing Challenge: From “Does it work?” to “Where does it work?”

Not long ago, the primary question around autonomy and electrification was simple:

Does the technology work?

In Mining, we now have a clear answer. Autonomous haulage is delivering consistently in live operations, at scale. Fleet-wide deployments from Komatsu and Caterpillar have removed any doubt - this is no longer a pilot phase.

Agriculture, in many ways, reached this point earlier, but in a different form.

Autonomous guidance, GPS-based steering, and precision farming systems have been in use for years - improving efficiency, reducing overlap, and enabling data-led decision making across large-scale operations.

The difference is that autonomy in Agriculture has typically been incremental, task-specific and integrated into existing workflows.

Whereas Mining has moved more aggressively toward full system-level autonomy.

Electrification is slightly different. The tech is working, but is highly dependent on the environment:

• Infrastructure availability

• Site layout

• Duty cycles

Which led to an initial conclusion (see my previous LinkedIn posts):

Autonomy is proven. Electrification is viable, but conditional.

The Constraint Shifts: Environment and Infrastructure

As the technology matures, a different constraint is rising to the forefront. The tech is no longer the challenge - the surrounding environment and infrastructure are becoming the limiting factors.

In Mining:

• Road conditions

• Dust

• Site design

• Power infrastructure

  
  

In Agriculture:

• Field variability

• Connectivity gaps (especially rural)

• Legacy equipment and mixed fleets

• Farm layouts not designed for automation

Autonomy systems don’t fail because the software is ineffective. Komatsu’s April milestone of commissioning 1,000 autonomous trucks demonstrates the technology works at scale. Failures arise when operating environments expose edge cases.

Electrification is at a different stage. It is not yet as advanced as autonomy, but its main challenge is not battery viability. The bigger issue is supporting infrastructure. Battery-powered haul trucks in mining and tractors in agriculture that pull very heavy implements still face technical limits, but the core battery technology is sound.

In many cases, especially in rural farming, operating context is the real constraint:

• Charging infrastructure (particularly in rural farming areas)

• Energy availability

• Operational constraints around uptime

Is Mining a Blueprint… or an Outlier?

Mining continues to push ahead:

• Autonomous fleets at scale

• Battery-electric haulage being deployed

• Integrated systems delivering productivity gains

  
  

This raises an important question. Are Mining, Agriculture, and Construction following the same trajectory - or are they fundamentally different problems? Because Mining operates in controlled environments with standardised workflows and predictable routes, whereas Ag & Construction operate in variable, less-predictable environments, with less standardised systems and more fragmented ownership.

This suggests mining may not just be ahead - it may be structurally advantaged.

Autonomous vs Electric OR Autonomous & Electric?

More recently, developments have moved the conversation forward again. Large-scale deployments are no longer just autonomous or electric; they are becoming integrated systems.

Autonomous, battery-electric systems operating with coordinated fleets, managed energy and optimised operations.

The key shift - It’s no longer about individual technologies, it’s about how they work together.

This becomes increasingly apparent when we look at recent industry moves:

1. Caterpillar acquiring Monarch Tractors (LinkedIn - CAT x Monarch)

2. Komatsu partnering with Applied Intuition – Komatsu and Applied Intuition partner

Caterpillar partnering with NVIDIA - Caterpillar | Caterpillar Teams with NVIDIA

Major OEMs are now investing heavily in:

• Software-defined platforms

• Embedded AI

• Continuous optimisation

• Acquiring capabilities (see my post on the Caterpillar purchase of Monarch Tractors)

These aren’t just incremental improvements; the OEMs are recognising the next challenge – what happens after deployment?

Because the systems we’ve seen to date share common characteristics - they are highly engineered, and highly optimized. What’s emerging now is different - machines (and the systems around them) are starting to adapt, learn and continuously improve, during operation, not just in periodically planned upgrades.

This changes where value sits. Previously, the key drivers of adoption were:

• Controlled environments

• Clear ROI

Now, rate of improvement is becoming equally important. Two identical fleets, operating in similar environments, may not deliver the same results. Not because of hardware differences, but because of how effectively their systems learn from data, optimise performance and adapt to conditions.

The constraint is shifting again. We must not only consider environment and infrastructure – but how quickly the system improves once it goes live.

A Brief Comparison: Automotive Autonomy

It is interesting to compare this with Automotive autonomy.

On-road autonomy has been heavily focused on solving edge cases, handling unpredictable environments and meeting regulatory requirements. As a result, progress has been slower, more complex and more constrained.

Off-Highway has taken a different route by operating in controlled environments with defined operational domains. Therefore, it has been able to deploy faster, scale earlier and deliver ROI sooner.

In many ways Off-Highway autonomy is ahead of Automotive in real-world deployment. But now both sectors are converging on a similar challenge - how systems improve over time through data and AI.

What This Means for the Off-Highway Sector

Looking across the sector today:

• Mining → autonomy is delivering, systems are maturing, and optimisation is underway

• Agriculture → technology is advancing, but the market remains fragmented

• Construction → progress continues, but environment and variability remain constraints

The shared theme is no longer adoption alone, but evolution.

Conclusion: From Machines to Evolving Systems

Across autonomy, electrification, and digitalisation, the pattern has been clear:

1. Prove the technology works

2. Deploy it in the right environment

3. Build systems around it

4. Optimise performance at scale

5. Continuously improve over time

We are now entering that final phase. That leads to a different question from the one we started with - are we still thinking in terms of autonomous machines or are we beginning to see the rise of autonomous systems that evolve over time?