From the Concept of Being “the Boss” to the Idea of Being “a Team”: The Adaptive Co-Pilot as the Enabler for a New Cooperative Framework

Round 1

Reviewer 1 Report

Authors present a method that driver and automation are regarded as members of a unique team. The topic is interesting and the authors present the method scientifically with a systematic way.

Some of my comments and questions are as follows. 

 

• The problem is specifically defined and presented in the introduction, and other existing studies and their shortcomings are well explained.
• Figure 1: It would be better if you friefly explain how it differs from the existing system since it is a description of the novel framework, though well aware of the details explained later. 
• Chapter 2.1: In the tables and equations listed from here, although common equations are included, it would be better that the parameter definitions be properly arranged. 
• Figure 6: As I am not familiar with it, but do the objects in the HMI assume detection or is it randomly placed?
• Table 6: In Shared Control, when the driver is attentive or distracted, the corresponding parameter values for different situations seem the same. What is the difference between parameters?
• The results are well analyzed and presented with appropriate figures and graphs. 

Author Response

Dear Editor and Reviewer,

We would like to thank you for the time and effort spent through the revision process, and especially for all the meaningful inputs and helpful feedback to improve the quality of the manuscript. The article was carefully revised following all the reviewer’s comments.

Best regards,

The authors

 

REVIEWER #1 COMMENTS

R1.1 - Figure 1: It would be better if you briefly explain how it differs from the existing system since it is a description of the novel framework, though well aware of the details explained later. 

Authors: We highly appreciate this suggestion, as there was no mention of the existing frameworks. In response to this comment, we have updated the introduction paragraph of Section 2 explaining the differences between common AVs frameworks (where automation is the Boss) and our proposed Team framework. (Linked to comment R1.1 in the manuscript).

Manuscript: This section presents the design and implementation of the adaptive Co-Pilot framework (shown in Figure 1). Standard frameworks where automation is the boss are commonly composed of six components (acquisition, perception, communication, decision, control, and actuation) [24]. However, when the human is considered a co-driver, new components are needed. First, the acquisition and perception modules are extended with the information provided by driver monitoring systems. Additionally, three relevant complements need to be defined when the driver is involved in the control loop: (1) a new control modality (Shared Controller), which is a lane-centering controller, with the ability to assist the driver with an adaptive level of authority (instead of fixed as in common AVs frameworks), either with lower or higher intensity, (2) a new decision-making component (Arbitration), to find out how much authority should be given to both the driver and automation, based on the driver-vehicle-environment context (particularly, the design for a use case of a distracted driver will be explained in this paper), and (3) a new interaction interface (Visual HMI) that complements the system aiding the driver through visual feedback increasing situational awareness and giving an indication on the current level of authority. The novelty of this framework relies also on the design considerations of each of these three components as explained in sections 2.1, 2.2, and 2.3 respectively.

R1.2 - Chapter 2.1: In the tables and equations listed from here, although common equations are included, it would be better that the parameter definitions be properly arranged. 

Authors: Dear reviewer, thank you for this comment. In response to it, we have added a footnote in Table 1, in order to define the meaning of some variables and parameters that were not explicitly defined through the text (Linked to comment R1.2 in the manuscript).

R1.3 - Figure 6: As I am not familiar with it, but do the objects in the HMI assume detection, or is it randomly placed?

Authors: We thank you for this relevant question. In this work, the HMI was presented as an initial prototype of the layout that sets the bases for the real implementation.  Therefore, it shows a realistic reconstruction of the simulated external environment, which will be linked to real-time data as part of our future work. A mention of this next step has been added as part of the future work, in the “Discussion” section. (Linked to comment R1.3 in the manuscript).

Manuscript: Integration of the HMI: while presented in this work as an initial prototype, the next step is to integrate the visual HMI with real data coming from the simulator, and to be used in the experimental test to evaluate its effectiveness as an enabler of trust in decisions and actions performed by the Co-Pilot.

R1.4 - Table 6: In Shared Control, when the driver is attentive or distracted, the corresponding parameter values for different situations seem the same. What is the difference between parameters?

Authors: This is a very valid comment, as in the Table both values appear as lambda = FIS. The intention was to inform that the authority value was calculated by the Fuzzy Inference System (FIS). In response to this comment, we have added a footnote to Table 7 differentiating between the two different conditions. (Linked to comment R1.4 in the manuscript).

Manuscript:
FIS_a Is the authority calculated by the FIS (arbitration module) when the driver is attentive
FISd Is the authority calculated by the FIS (arbitration module) when the driver is distracted

Author Response File: Author Response.docx

Reviewer 2 Report

The thought of “The Boss” and “A Team” principle in terms of highly-automated vehicles is very significanr and timely. The paper presents the adaptive Co-Pilot cooperative framework of a Team, which both includes human and artificial system forms. The paper details the design principles of the shared controller of the adaptive Co-Pilot.

The optimization takes into account tracking performances as well as performances related to driving comfort, driving safety and driver effort. It also develops and demonstrates a system of constraints used in the optimization procedure. The optimization uses both arbitration and shared algorithms. The paper is easy to read and enjoyable. However, the number of abbreviations is extremely large. For this reason, I recommend the use of a list of abbreviations. The rules formulated in fuzzy terminology can be well followed. A question is arisen: what is the theoretical explanation for choosing such membership functions (page 11)? The design method applies three main constraints, i.e., control input, lateral error, and control torque, in experimental tests, which apply to. The choice of of the FIS for ‘\lambda_dim’ is hardly understood (see Table 6).

According to the reviewer, the paper contains so many valuable thoughts that it would be worth sharing with the MDPI readership. However, a lot of questions have also arisen. Some of them are the following: What were the considerations for choosing KPIs? What were the considerations for choosing the constraints? What control design algorithm should be applied? If it is a type of fuzzy design, it should be justified. How the performances and robustness can be guaranteed? How the feasibility can be achieved?

Author Response

Dear Editor and Reviewer,

We would like to thank you for the time and effort spent through the revision process, and especially for all the meaningful inputs and helpful feedback to improve the quality of the manuscript. The article was carefully revised following all the reviewer’s comments.

 

Best regards,

 

The authors

 

REVIEWER #2

R2.1 - The paper is easy to read and enjoyable. However, the number of abbreviations is extremely large. For this reason, I recommend the use of a list of abbreviations.

Authors: Thank you for this valuable comment, we are glad to know that It was an enjoyable article to read. To facilitate the reading, we have followed your suggestion and added an abbreviation list as Table 1, located at the beginning of the introduction (as most of the abbreviations were in the first pages of the article). (Linked to comment R2.1 in the manuscript). Additionally, the corresponding abbreviations were removed from the text.

R2.2 - A question is arisen: what is the theoretical explanation for choosing such membership functions (page 11)?

Authors: We highly appreciate this relevant question. The choosing of the membership functions followed two steps. First, some representative values related to the use case were chosen. Secondly, experimental sessions helped to adjust the MFs to achieve the desired behavior of the decision surface (Figure 5). The corresponding paragraph of Section 2.2 was updated accordingly, explaining the motivation of the most relevant values of the MFs for the inputs and output of the system. (Linked to comment R2.2 in the manuscript)

Manuscript: Concerning the design of the MFs, the values for each label are representative rather than exact values (this is one of the advantages of FIS algorithms, as it resembles how humans define variables). In terms of lateral error, 1.5 m (HIGH) is the distance when the vehicle is at the lane border, while 0.3 m (NONE-LOW) was selected to give the driver some freedom to deviate from the lane center without feeling intervened by automation. On the other hand, the MFs for driver distraction were selected by observing the raw distraction signal of different drivers. With respect to the output (λ_dim), 15 Nm (HIGH) is the maximum steering motor torque, while the other values were selected experimentally based on subjective driver feeling on the steering: a) MED, strong but possible to override (10 Nm), b) MED-LOW, strong guidance with some driver freedom to move (6 Nm), and c) LOW-MAN, soft guidance barely felt at the steering (2 Nm).

R2.3 - The choice of the FIS for λ_dim  is hardly understood (see Table 6).

Authors: This is a very reasonable comment as it was not explained clearly through the text. Therefore, we added a footnote to Table 7 to reinforce the meaning of the authority value, with a reference to the corresponding section and equation. (Linked to comment R2.3 in the manuscript).

Manuscript:
λ_dim: Is the control authority given to the automation (steering wheel stiffness), and is also defined as the maximum torque applied by automation (reached when the vehicle is at the lane border) thanks to the following link with the control torque constraint |Tmpc| <= λ_dim  (refer to Section 2.1.5 and Eq. (13))

R2.4 - What were the considerations for choosing KPIs?

Authors: Dear reviewer we really appreciate this question. We have included a footnote on the KPIs section (3.1.1) explaining this in detail. (Linked to comment R2.4 in the manuscript).

Manuscript: KPIs have been selected in the context of the PRYSTINE project [53], by conducting specific project workshops with experts in the field. Previous works in shared control studies [8] were also used as a reference to select the representative measurements of the main fields of evaluation.  

R2.5 - What were the considerations for choosing the constraints?

Authors: We thank you for raising this question, as the motivation for constraints design should explicitly be indicated. We have updated the corresponding text in section 2.1.2. (Linked to comment R2.5 in the manuscript).

Manuscript: Moreover, safety considerations are added by applying a constraint on the yaw rate (ψ), to avoid vehicle to unsafely drift (tested experimentally, and coherent with a related work [37]), and also limiting the maximum allowed lateral deviation (defined approximately as half of the lane width, to avoid vehicle to depart from the lane). Additional constraints are added to limit the steering wheel behavior (θ, w, Tmpc, ΔTmpc) based on the subjective feelings of drivers in the simulator.

R2.6 - What control design algorithm should be applied? If it is a type of fuzzy design, it should be justified. How the performances and robustness can be guaranteed? How the feasibility can be achieved?

Authors: Thank you very much for this comment. In our design, the fuzzy logic serves the purpose of a decision-making system (context supervision), instead of a controller. In this sense, the benefit of the design we presented is that the decision-making should not handle the performance, robustness, and feasibility, its main task is to calculate the proper level of authority according to the context. Instead, the shared controller is the one that applies the calculated authority while keeping the performance, as demonstrated in 2.1.4 and Figures 3 and 4, using the well-known method of NMPC. We have added an additional paragraph at the end of Section 2.1.5 to highlight this benefit of the system. (Linked to comment R2.6 in the manuscript).

Manuscript: With this in mind, one of the benefits of the proposed framework, is that as the low-level controller based on NMPC can keep the performance, robustness, and feasibility independently of the authority level, then, the high-level decision-making (the arbitration system) is given more flexibility and can be adapted for different use cases without the risk of modifying the fundamental controller behavior.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors provided correct answers to my questions and the paper was amended accordingly.