Pattern recognition in Robot Vacuum Map

Hello,

I am currently working on the recognition of fallen persons from a generated map of a vacuum cleaner robot (2D lidar sensor). This should be kept as simple as possible in the first step.

In the appendix are the pictures of an empty room as well as of a room with a person on the floor. I would be happy if someone could guide me in one direction how I could do this detection with Matlab.

Thank you very much and best regards

Michael

Answer 1

[Sivsankar]

Hi Michael,

You can leverage this MATLAB answer as an initial direction to start:

https://www.mathworks.com/matlabcentral/answers/447374-pattern-recognition-pattern-comparison?s_tid=answers_rc1-2_p2_MLT

I can understand that this answer might be vague in your case since you have used 2D LiDAR dataset. I assume that you are having unsupervised 2D LiDAR data. So, you could use choose some prominent pattern recognition algorithms such as K-Means Clustering which is suitable for unsupervised learning. However, keep in mind to execute necessary preprocessing steps before employing your algorithm onto the dataset. You can refer to this documentation to get a better understanding in the algorithm:

https://www.mathworks.com/help/stats/kmeans.html?searchHighlight=K%20means%20clustering&s_tid=srchtitle_support_results_1_K%20means%20clustering

There is also a dedicated LiDAR toolbox through which you could detect objects. You can learn more about the toolbox from here:

https://www.mathworks.com/help/lidar/segmentdetectlabel.html

I believe that this information would suffice as a starting point for your pattern detection algorithm in MATLAB.

Thanks.

Answer 2

[Image Analyst]

See if this might help:

https://www.mathworks.com/help/vision/ug/abandoned-object-detection.html?s_tid=srchtitle_site_search_1_abandoned%20object