The greatest challenge of the project was the camera module. The biggest hurdle initially was deciphering the arducam sample code and its libraries in order to determine how the jpeg data was being captured, recorded and then sent to an external html file to be displayed. After countless office hours and several days, we were able to pull the jpeg data from the camera directly. We then had issues with the images being incomplete or poor quality so we had to set up a system within the existing code which would enable us to collect the image in its entirety before sending to the server rather than piece by piece as the original code was set up to do. Another issue that arose from working with the camera was when we attempted to send the base 64 encoded jpeg data to the server from the ESP32 it wouldn't display the image and the file sizes including the images would vary despite the images being the same sizes. In the two days it took to figure out that the data should be sent using JSON rather than url encoded, there was a great deal of scrutinizing identical strings that were interpreted differently and magically seemed to increase in storage size. Another difficulty was dealing with the MP3 modules. Having realized the MP3 module that was working successfully for a week was suddenly heating up, we decided to swap it out for a brand new one, and, when the issues persisted, agreed that the problem must be with the other components. We proceeded to swap all the hardware, including the ESP32, the wires on the breadboard, the breadboard itself, the cable connecting the ESP32 to the laptop, and even the laptops. It was discovered that the MP3 modules were knock-offs, and once we were able to attain a new one, it worked perfectly. It is believed that when she heard the news, LASSIE actually shed a tear.
In terms of future work for this robot, we would like to integrate more cameras and ultrasonic distance sensors, in order to attain a higher level of object avoidance. We would also consider adding lights, night vision cameras, and gas sensors, as well as a microphone, to gain more information from the robot. Aside from increasing the robustness of the robot, as well as how long it can run on each charge, any future work we did would most likely be to integrate more informational sensors into the robot, in order to gather as much information as possible. Another idea we had was to shift the focus of the project- instead of making LASSIE an all in one robot, make it an accessory, something that can be attached to any range of autonomous vehicles (quadcopters, rovers, etc) in order to gather more information with a much higher amount of flexibility. We believe that our work laid the foundation of what a cost-efficient first responder robot could be like, and built on it by adding a few extra features as well.