Obstacle Avoidance Trajectory Planning for Autonomous Vehicles on Urban Roads Based on Gaussian Pseudo-Spectral Method

Round 1

Reviewer 1 Report

The article entitled "Obstacle Avoidance Trajectory Planning for Autonomous Vehicles on Urban Roads Based on Gaussian Pseudo-spectral Method", is interesting current, well well-structured. The scientific contribution is highlighted.

Certain parts of the article need to be explained in more detail.

I asked the authors to explain why the city road was chosen. The transverse profile of 2x3 lanes can suit any high-performance road (e.g. suburban highway) and is not a typical city road. Only vehicle-vehicle conflicts were analyzed, which is also not characteristic of city roads.

Line 52: Why is RRT* with an asterisk and no additional explanation?

Line 77: What is a PID controller?

Line 124: What is the Hybrid A* and why does it have an asterisk?

In mathematical expressions (1) explain all variables, and then in expressions (2)-(4) explain all new variables.

Explain variables in all mathematical expressions when they first appear.

In Lines 207-214, the assumptions of the model are written. In the further development of the model, could the longitudinal slope of the road be the input data of the model? This would bring the model closer to real conditions, but it would affect the modeling results. I ask the authors to comment on it.

Figure 4 shows the analysis of the vehicle-vehicle collision in the conflict zone of the intersection. It can also correspond to the entrance lane of any highway. What specifics indicate that it is a city road when no other conflict was analyzed except vehicle-vehicle, which is not characteristic of a city intersection?

The schematic representations in Figures 9 and 10 are explained in the text. They are not mentioned or explained in the text at all. In case it is not necessary to refer to them, it is questionable why they are in the article.

Why was the SQP sequential quadratic programming algorithm chosen?

Chapter 4 shows the results of simulating the lane change of vehicles located in the middle lane. At the same time, the vehicle changing lane accelerates, while the overtaken vehicle drives at a constant speed. Does the model take into consideration vehicles in the target lane? I ask the authors to comment on it.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

I appreciate the focus of the study of urban AV on city roads which is extremely challenging due to various constraints while lane-changing scenarios were taken as a basis. The approach to integrate dynamic vehicle constraints and changing boundary conditions into the optimization problem's constraint set including environmental obstacle and collision avoidance and developing a performance metric with time and turning angle as parameters deem to be innovative. The set of methods employed, notably GSM to discretize the state and control variables as well as to compare its computational efficiency with LPM and DSSSM (for two different discrete interpolation points), is impressive and valuable. The optimal control model for local path planning has been tested against experiments with two weights, the results demonstrate its stable and computational efficient performance and ability to guarantee safe obstacles avoidance. In my view the paper is well structured and written (a few minor flaws, e.g., see line 137 to 138, misprints and layout inconsistencies should be corrected). The state-of-the-art and problem statement are well developed although the list of references appears slightly outdated.  The limitations of the current methods (GPM) are  well documented and future work is clearly outlined.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper has been improved with additional explanations of the methodology. The work is interesting and can serve as a basis for continued research by the same or other scientists, so it is necessary to clearly describe every limitation of the methodology.

Reviewer 2 Report

Thank you for your response and correction of printing errors and layout as well as additional explanation. I am more than happy to recommend the acceptance of your revised article.