‘Does Dutch Collision Law Require Reinvention for Maritime Autonomous Surface Ships (MASS)?’ by Raghini A. Acharju

Introduction: Administrative Approval Meets Fault-Based Liability

Dutch inland collision law appears sufficiently flexible to accommodate automation, as it already allows fault attribution through both human error and vessel defects. The increasing use of automated navigation systems in inland shipping therefore may not necessarily require structural reform of the existing liability regime.

Automated navigation systems are increasingly used in Dutch inland shipping. Systems such as Trackpilot assist in route planning, course-keeping and traffic anticipation while the master remains on board and retains override capability. From the perspective of Dutch authorities, these developments are primarily approached through administrative oversight: certification, operational limitations and safety supervision.

This article primarily focuses on automated navigation systems operating within a supervised environment, such as Trackpilot-type systems, where the master remains onboard and retains override capability. References to higher levels of autonomy and MASS are included only to assess whether the same doctrinal reasoning would remain applicable in more autonomous operational settings.

However, when a collision occurs, the dispute will not be decided under administrative law. It will be resolved under Dutch inland collision law, which is fundamentally fault-based. Automation does not appear to alter that fault-based structure; it primarily alters the evidentiary and factual complexity of litigation.

This creates a practical tension. Administrative certification authorises the use of automated systems. Yet liability in collision cases depends on fault attribution, causation and apportionment under private law.

This article examines what the certification and use of automated navigation systems mean in practice for collision disputes under Dutch inland law. The central question is:

“To what extent does Dutch administrative certification and use of automated navigation systems affect the attribution of fault in inland collision disputes?”

This article argues that, on the basis of current Dutch collision law and particularly the broad interpretation of fault adopted in Casuele/De Toekomst^[1], the existing liability framework appears capable of accommodating foreseeable forms of maritime automation without immediate structural reform. However, this should not be understood as excluding future legislative intervention; rather, it suggests that, under current doctrine, automation-related challenges primarily concern evidentiary complexity and fault attribution rather than the fundamental structure of Dutch collision liability law.

The Dutch Collision Liability Framework: Doctrinal Breadth and Flexibility

Dutch inland collision law is based on fault. Liability depends on blame attributable to a vessel. Where two vessels are at fault, liability is apportioned proportionally.

Three practical features define the regime:

Fault as the basis of liability
There is no strict liability merely because a vessel was involved in a collision. Fault must be established.
Flexible apportionment
Courts allocate responsibility according to the degree of blame, allowing nuanced factual assessment.
Fact-intensive adjudication
Collision cases depend heavily on reconstruction through logbooks, AIS data, radar images and expert evidence.

Crucially, Dutch Supreme Court case law suggests that “fault of the vessel” is interpreted broadly. In Casuele/De Toekomst, the Dutch Supreme Court held that a vessel may be considered at fault not only where damage results from:

• navigational error by the shipowner, master or crew
• persons working for the vessel
• an inherent defect of the vessel itself

The significance of this judgment lies in its suggestion that fault attribution under Dutch collision law is not limited to human navigational error, but may also extend to technical deficiencies attributable to the vessel.

The concept of fault is therefore not limited to human navigational error. It may also encompass technical deficiencies attributable to the vessel. A malfunctioning steering mechanism, defective radar or flawed onboard system may qualify as a defect of the vessel giving rise to fault.

The Dutch collision liability framework therefore appears structurally flexible. By covering both personal error and inherent defects, it may provide sufficient conceptual capacity to address incidents involving automated navigation systems.

Technological novelty does not automatically entail doctrinal novelty.

There appears to be no immediate structural need to reinvent the Dutch liability regime for MASS. The existing framework appears capable of accommodating foreseeable technological developments.

Trackpilot and the Question of Control

Trackpilot-type systems operate within a supervised model. The master remains on board, retains override capability and is expected to monitor the system.

From a liability perspective, collision litigation involving automated systems will revolve around the question of control.

Four practical issues are likely to dominate proceedings:

Active supervision
Was the master adequately monitoring the system? Certification presupposes human oversight. Failure to supervise may constitute fault.
Timely intervention
Was intervention required? Failure to override in time may amount to negligent navigation.
Compliance with certification conditions
Non-compliance with operational limitations may weigh heavily in fault analysis.
Crew competence
Insufficient understanding of the system may support a finding of fault.

In practice, automation shifts the litigation inquiry from:

“Who physically steered the vessel?”

to:

“Who exercised effective control over the navigational decision-making process?”

For supervised automation systems, the master appears to remain the primary norm addressee. Automation does not eliminate responsibility; it reframes the context in which it is assessed.

The “reasonable master” standard is therefore likely to evolve.

System Malfunction: Error or Defect?

Automated navigation introduces an additional layer to collision analysis: system malfunction.

Under the principles confirmed in Casuele/De Toekomst, fault may arise through an inherent defect of the vessel. An onboard automated navigation system, once integrated into the vessel’s operational structure, may arguably fall within this category.

Two analytical routes are therefore possible:

Negligent supervision route
The system functioned within parameters, but the master failed to monitor or intervene appropriately.
Defect route
The system malfunctioned or was inherently defective.

This dual pathway suggests that Dutch collision law may not require doctrinal expansion to accommodate automation. Existing doctrine appears capable of addressing both human error and technical malfunction.

What changes is not the legal structure, but the factual investigation required to determine where responsibility lies.

Administrative Certification: Shield or Strategic Factor?

Administrative certification does not create civil immunity. A certified system may still be involved in negligent navigation or may arguably qualify as a defective component of the vessel.

Certification is therefore not legally irrelevant, but it does not eliminate responsibility.

Higher Levels of Autonomy: Does the Analysis Still Hold?

A potential counterargument is that increasing levels of autonomy may render fault-based liability more difficult to apply.

However, Dutch collision law appears capable of accommodating certain non-human sources of fault.

Even at higher levels of autonomy, liability may arguably still be analysed through:

• defects in system design or performance
• failures in system oversight at organisational level
• inadequate operational frameworks

While higher autonomy may complicate factual analysis, it does not necessarily follow that structural reform is required.

What Will Likely Change — and What Will Not

What will not change:

The fault-based nature of Dutch collision law
The central role of the master as norm addressee
The possibility of attributing fault through inherent vessel defects

What will change:

The complexity of factual reconstruction
The evidentiary importance of technical data
The frequency of multi-party disputes

Conclusion

The introduction of automated navigation systems such as Trackpilot does not appear fundamentally to alter the structure of Dutch inland collision law.

The regime remains fault-based and appears doctrinally flexible. It accommodates both personal navigational error and defects inherent to the vessel, as recognised in Supreme Court jurisprudence.

While automation introduces new factual and evidentiary challenges, current doctrine does not appear to require immediate structural reform. Collision disputes are likely to focus increasingly on supervision, system integrity and technical evidence.

Automation does not eliminate responsibility. Rather, it redistributes and complicates responsibility within an existing legal framework that appears capable of adapting to technological change.

This conclusion should not be understood as excluding future legislative intervention. Rather, it suggests that, under current Dutch doctrine and with respect to presently foreseeable forms of supervised maritime automation, the existing collision liability framework appears capable of accommodating technological developments.

^[1] Casuele v De Toekomst (Supreme Court of the Netherlands, 30 November 2001) NJ 2002/143.