Machine learning will be used for object detection. Throughout the development of this project, there have been multiple compatibility issues with the Raspberry Pi 4, ROS Noetic, and TensorFlow. This problem has been addressed and experimentation with a new microprocessor will begin during Semester II. The selected microprocessor is the NVIDIA Jetson Nano Development Kit. The Jetson Nano is more compatible with open-source software. Once TensorFlow is installed we will begin training the neural network. When the neural network is verified we will be able to determine the percentage of object identification accuracy. A fabricated example of these results can be seen in Figure 18. Additional difficulties were faced with the integration of the Arducam Mini Camera Module. Compatibility issues were addressed and the camera was enabled upon system boot within the system configuration file. The step to solving this problem can be found in Appendix C.
