Ontology of autonomous driving based on the SAE J3016 standard

This paper examines the important problem of formalize vehicle
standards. In particular, the authors formalize the SAE-J3016 standard
of levels of autonomous. The paper is original and significant,
although the writing and descriptions of the resources could be improved
for better readability (using examples).

*Originality* The authors have demonstrated their creativity in
constructing an ontology that captures the SAE-J3016 standard. The
ontology is explained thoroughly and in detail throughout the paper.
Their work opens a new area towards a more comprehensive understanding
of AV-related concepts and organization.

*Significance*: The work is significant as it advances the knowledge in
the AV domain and it sets a new standard for future research in
formalizing vehicle standards.

*Quality of Writing*: The writing of the paper is moderately clear,
although there is room for improvement. The logical structure and flow
of ideas make it easy to follow the concepts and definitions that are
presented. However, the paper and the quality of the writing could be
strengthened by incorporating a running example or set of examples to
illustrate the ontology definitions. For example, the figures of the
ontology structure is a great illustration, however the captions could
be improved to describe what the part of the ontology is being
described with an example.

For example, the authors could include the example (Section 7.3) earlier
in the paper to illustrate the ontology on a real-world example. A
further benefit is that including examples could motivate the broader
impacts and implications of their work. Regardless, this paper is
well-written, but sometimes the ontology formality and semantics is
difficult to understand. With a bit more descriptive text, this paper
has the potential to become an even more valuable resource.

*Artifacts*: The artifacts are complete. There is a github link.

*Specific questions and improvements*:
- The motivation of the problem in the introduction and in the previous
work could be better explained. For example, one motivation is the
problem of different nominclature of the different levels of autonomy,
which was not fully highlighted.
- The paper coule be strengthened with an example of where the ontology
would be useful, possibly a running example that could be included
throughout the full paper, e.g., V2V communication.
- The authors mention other standards from other institutions, and I was
wondering if there are geographic differences between these standards.
For example, are the US based NHTSA policies different from the BSI
standard? Would the contributed ontology unify these ideals.
- The yellow in Figure 1-3 is a bit harsh to see, perhaps muting this
color would be help. Also, I think the caption could be expanded to
describe the key takeaway of the BFO taxonomy fragment.
- I'm wondering if the roles could designate blame in the case of
failures or accidents. This may be a topic to elaborate on in the
conclusion or in a discussion section.
- The first sentence on page 5 line 22, is a bit difficult to
understand. In fact, the paragraph might be strengthened by adding an
example of a disposition.
- What is an example of the "intuitive notion of normal condition" (page
9 line 22). Does this correspond to a well-paved road and "normal"
weather?
- I'm wondering about the differentiation between the HumanUserRole and
the HumanUser. This is a key point that may want to be elaborated on.
- The example: section 7.3 might want to be included in the paper
earlier.
- Validation of the ontology.
- The paper may be strengthened with a summary of the SAE-J3016, perhaps
including an infographic, e.g.,

This paper introduces SAE J3016 standard based ontology for autonomous driving, which is based on BFO ontology. SAE J3016 standard provides the taxonomy and definitions for terms related to driving automation systems for On-Road Vehicle, describing six levels of driving automation. This ontology describes motor vehicles, vehicle systems, driving tasks, capability, roles of persons, and autonomy levels, etc in Description Logic. Ontologies are accessible via https://github.com/kul-ai/ontology-autonomous-driving.

Detailed comments are as follows:

Some statements are difficult to understand. E.g.
Page 5. Realization of a disposition occurs when and because its bearer is in some special physical circumstances, but this realization is strongly based on their physical makeup.

Page 7:
Although ADS is mentioned in Section 2, it would be great to mention the full term Automated Driving Systems (ADS), rather than just saying ADS-equipped in Section 5.

Is ADSEquippedDualModeVehicle a subclass of ConventionalVehicle? Fig 2. Shows there is isa relation between them, and it’s a ADSDedicatedVehicle. But in the ontology, ADSDedicatedVechile and ADSEquippedDualModeVehicle are subClasses of ADS-equipped vehicles. Fig 2. Is not consistent with the ontology hierarchy.

Page 12. The definitions/naming of DynamicDrivingTask(DDT) and DynamicDrivingSubtask(DDS) are confusing (e.g. definition (37) , (109)). From the naming, it looks like DDS is part of DDT, but DDT is_a capability, and DDS is_a function. It might be better to add “Capability” or “Function” as suffixes to avoid confusion of their definitions.

What is the difference between Driving automation systems (DAS) and Automated Driving Systems (ADS)? Both terms appear frequently in the paper.
On page 11, Isn’t Full-trip feature a DAS feature, rather than ADS?

7.3 Example section is difficult to follow. It would be easier to follow the instances, if the authors can use the class info (e.g. i3_Vehicle, i1_ADS, i6_HumanDriver, etc), instead of just using i1, i2 for naming the instances. What are the instances of i8 and i10 in Fig. 5? If possible, please use a bigger font size for the figures.

Fig.6, adding labels of the 5 levels in the figure will better help readers understand the hierarchy.

The definitions of (124) and (125) are confusing. What’s the differences of these two definitions? Can we make it into one?

Can we combine (132) and (133) ?

Page 25, is there any relationship between Trip and SubTripFeature?

Reference: Please check the following paper, and if possible, compare with your ontology.
Ontology and Lexicon for Automated Driving System (ADS)-Operated Vehicle Behaviors and Maneuvers in Routine/Normal Operating Scenarios
https://www.sae.org/standards/content/j3164_202301/

Summary:
The paper proposes the development of an ontology using formal ontology methods to enhance the understanding of autonomous driving and the SAE J3016 standard. The authors ground their work on a top-level BFO ontology and provide a formal account of various areas covered by the standard, including motor vehicles, driving tasks, roles of persons in road communication, and autonomy levels. However, the paper lacks a evaluation or validation of the developed ontology, which could raise questions about its effectiveness and applicability in real-world scenarios . Future work suggested in the paper includes the need for further validation and evaluation of the ontology to assess its performance, compatibility with existing systems, and potential for interoperability. However, I believe it is crucial to incorporate the evaluation in this work.

Originality:
-The originality of the paper lies in its application of formal ontology methods to enhance the understanding and clarity of the SAE J3016 standard and its implications for autonomous driving.

Significance of the results:
-The results of the paper provide a formal ontology of autonomous driving based on the SAE J3016 standard, which contributes to a better understanding of the concepts and relationships within the field.
-The ontology also provides insights into the process of driving automation, covering areas such as motor vehicles and their systems, driving tasks and subtasks, roles of persons in road communication, and autonomy levels.
-The significance of the results lies in the potential application of the ontology in enhancing the comprehension and development of autonomous driving technologies, as well as facilitating interoperability with existing information systems.

Quality of writing:
-The paper is well-written and structured!

Strong points:
- The ontology is available on github https://github.com/kul-ai/ontology-autonomous-driving
- The ontology is in an expressive DL variant, SRIQ(D).
- reused BFO ontology, and include labels and other metadata.

Weak points:
- Did not mention ontology development methods, is it Methontology?
- Did not include the evaluation of the developed ontology, e.g., answering competency questions or use-case-based evaluation?

Overall, I believe that the paper is contributing to the state-of-the-art and, with the weak points addressed, would be a good fit for this journal.