We started our week by reflecting on the proposal presentation and feedback. It was clear to us that we need to refine our requirements based off of the target user experience, rather than the specs of each individual component. There was also some confusion over the scope of the project, so we’ll need to make this clearer next time in the design presentation. 

After the first proposal presentation, our group had some concerns about the processing power needed to run SLAM, read sensors, and do CV. After recommendations from Tamal and Joe, and further discussion, the team decided to use a spinning RPLidar to gather 2D cloud point information, instead of the RealSense camera. The RPLidar formats the point information in a convenient format to process, and seems to require lighter processing time than the RealSense. If the processing becomes a major issue later, we can also consider using two Pi’s to distribute the load, and have them use ROS to communicate.

This week, we also looked at each part in more detail to ensure compatibility, with regards to port connections as well as power. Then, we drew out an updated block diagram to confirm that all connections would fit onto one Pi. Ultimately, this is what our project will look like:

*insert block diagram*

View post on imgur.com

This coming week, we will finalize our design presentation, and pick up our parts from Quinn. With the parts, we will also begin setup and some preliminary testing. We want to find any baseline issues immediately, so that we can have maximum time to make changes. Also, we can return the parts as early as possible so someone else can use them.