Toyota RAV-4 Sticker and Hole Automation Algorithm

Development Team: Camron Sabahi-Pourkashani, Joey Maillette, Jeffrey Luo, Shahmeer Ali

Introduction

This proposal outlines a reliable solution for utilizing automation to detect and inspect holes in engine bays that need to be covered by stickers. The stickers are placed to reduce cabin-noise allowing for a smoother driving experience and prevent damage to the internal systems in the engine bay. The objective is to develop a program capable of analyzing a live feed/photos of an engine bay during the sticker application process, automatically identifying any holes that are not covered and relaying that information to the already autmated process.

Solution Overview:

To achieve the desired automation, we propose the following solution components and workflow:

Live Feed Acquisition:

Set up a high-resolution camera or a network of cameras to capture a live feed of the engine bay during the sticker application process. The camera(s) should be strategically positioned to provide optimal coverage of the entire engine bay area.

Image Processing and Analysis:

1. Preprocessing: Apply image preprocessing techniques to enhance the image quality and remove any noise or artifacts that may interfere with subsequent analysis steps. This involved operations such as noise reduction, image enhancement, and calibration.

2. Hole Detection: Apply computer vision techniques to identify potential hole locations. This involved edge detection, contour analysis, blob detection, and mean adaptive thresholding methods to locate areas with missing stickers, which indicates the presence of holes.

3. Precision Measurement: Calculate the size of detected holes by measuring their dimensions accurately. This can was implemented to differentiate holes that are apart of the engine and intended to be there, versus holes that need to be covered by a sticker.

Implementation and Integration:

1. Software Development: Develop a robust software application that integrates the various components mentioned above into a cohesive system.

2. Hardware Integration: Establish a seamless connection between the live feed acquisition system and the software application for real-time data transfer and analysis.

3. Testing and Optimization: Conduct extensive testing to ensure the accuracy, reliability, and speed of the system. Optimize the algorithms and parameters as necessary to achieve the desired performance.

A Next Step

After the algorithms previously mentioned detect a hole and ensure that hole must be covered with a sticker, we additionally created a machine-learning model to evaluate the precision and quality of the sticker placement. This was implemented by creating a CNN (convolutional neural-network) that was able to detect wrinkles after the placement and training the model to identify whether or not the machine correctly placed this sticker. Although we did not have the data or resources to implement this advancement fully, this serves as the grounds to further improve this automation process.

Built With

In order to develop this computer vision algorithm, we have utilized various tecnologies and frameworks:

• 
• 
• 
• 
• 
• 

Conclusion

By implementing the proposed solution, which combines advanced image processing techniques, and object detection via computer vision, we can achieve reliable automation for detecting and inspecting holes in engine bays that must be covered by stickers. This solution will enhance quality control processes and ensure the precise application of stickers within the specified tolerances.