An Artificial Neural Network Memorizing Driving Paths

Step 1. Run Program10Z.exe

Step 2. Click the ‘Create Path’ button ( ③). The following map will be displayed. On the map, click 24 points in the order that a driver would pass. It is assumed that between successive two points, the car moves in a straight line.

Step 3. After the 25th point is clicked, the data points for the path will be stored while the first image will be displayed again with the map image disappeared. The stored path points are displayed on ‘Input data for the Neural Network’ section of the image (①). An input data for the neural network is partial path beginning from the first point to kth point where 0 < k < 24 and the output is the (k+1)th point ( ② ). 

Step 4. Click ‘No. Iterations’ button (②) to set it to 400,000 or a larger number. In case a smaller number of iterations is used in the training of the neural network, generated path may be off from the original path at some points.

Step 5. Click ‘Start Learning’ button (③) to start the training of the artificial neural network.

Step 6. Click ‘Learned Path’ button (⑤) to see how the path is learned. It should look the same as the original input path.

KamelConsulting, LLC.

CEO. Byung S. Moon | Tel. +82-10-6627-4306 | ㅣ Biz License 314-86-26086

2F, DICIA ,20 Daedeok-daero 512, Daejeon, S. Korea

2022 Kamel Consulting, LLC