An Angular Acceleration Based Looming Detector for Moving UAVs

Round 1

Reviewer 1 Report

This paper discusses role of visual perception in unmanned aerial vehicles and introduces A-LGMD model to address challenges in real-time obstacle avoidance. The paper proposes incorporating higher-order information on angular size, revealing a linear relationship between peak times of such information and time to collision. The paper also suggests that incorporating multiple visual field angle characteristics can enhance the model's resistance to motion interference. The methods such as "Enhancing assessment of corn growth performance using unmanned aerial vehicles (UAVs) and deep learning","Fusion and Enhancement Techniques for Processing of Multispectral Images", may also be discussed in the paper. The experimental results on synthetic and real-world datasets demonstrate the model's efficiency in detecting image angular acceleration, filtering out background motion, and providing early warnings.  The paper lacks mathematical support to the proposed method. relevant mathematical equations may be included.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper introduces the A-LGMD, a concise model for detecting looming objects, specifically addressing issues related to imprecise feature extraction and inadequate time-series modeling during rapid UAV self-motion. The primary aim is to improve the model's efficacy in swiftly and accurately detecting looming objects, drawing inspiration from the LGMD, known for its heightened sensitivity to acceleration information. While the paper is well-motivated, several areas could benefit from further attention:

1. The abstract appears somewhat lengthy; consider shortening it by emphasizing why the angular acceleration-based looming detector is proposed.

2. In Figure 2, provide an explanation of the implication of Izhikevich for better understanding.

3. Improve the quality of Figures 3 and 5, ensuring that the captions are clear and easy to follow.

4. Conduct a recent contrast analysis with papers such as references 10.1016/j.ijcce.2021.11.005 and 10.1016/j.ijcce.2022.07.001.

5. List all abbreviations in a table for clarity and explicitly state the study limitations of the A-LGMD.

6. In the last section, present more detailed future topics, especially those related to large models.

7. Enhance the overall presentation quality and language throughout the paper for improved clarity and coherence.

Minor editing of English language required.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper presents a model based on image angular acceleration for detecting looming objects. The model intends to addresses problems related to imprecise feature extraction and insufficient time-series modeling during the rapid self-motion of unmanned aerial vehicles (UAVs). However, the technique is not clearly presented.

The abstract is too long with a lot of background information. It should be shortened to summarise the key works in the paper.

All figures and tables should be discussed or mentioned in the main text. Quite a few figures are not mentioned in the main text. There are a few places in the main text where Figure (x) or Table (x) are mentioned.

In the second page, the caption of Figure 1 mentions Eq.A1, Eq.A2, and Eq.A3. Eq.A3 is also mentioned in the text. However, there are no such equations in the paper.

What is the meaning of Eq(1)? Eq(1) basically indicates that P(x,y,t)=|L(x,y,t)-L(x,y,t-1)|. The definitions of L and P should be clarified and, in particular, what are the differences between “over time” and “at time”?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

The authors have adequately addressed my comments and the revised manuscript can be accepted.