This week, I spent most of my time working on building a robot base with Hirani, ordering additional parts, and researching Simultaneous Localization and Mapping (SLAM).
During building sessions, we cut aluminium extrusions from metal shop at TechSpark after finalizing dimensions of our robot. Then, we waited a few days for connectors to come in and used them to connect all the extrusions we prepared. Also, I quickly learned how to use the laser cutting machine and cut out acrylic boards for the sides of the robot. It took about 6 to 8 hours to finish the frame.
I’ve also looked at materials for the conveyor belt. I was still debating on how I should mount the conveyor belt on the aluminium extrusions. But, Hirani and I during the build agreed upon using only two pvc pipes to wrap around the belt and attach the pipes to the motor that is attached to the extrusions. This solution seemed to be simple but solid enough for our pick up mechanism.
Meanwhile, Hirani and I worked on path finding algorithm and movement algorithm. As I’m working with path finding algorithm mostly, I confirmed with professor Mario and Aidan during our weekly meeting to use Apriltags to set up the boundary so that the robot would detect it before it begins to clean. Our prior concern was running Issac ROS on Jetson GPU might affect the speed of machine learning model training. Fortunately, as there were some prior projects built up on Raspberry pi, I decided to switch to Rpi to run Issac ROS. This way, as I’m working with Hirani most of the time, it would be more convenient for us to communicate how we are going to implement path finding algorithm and movement algorithm.
Also, I was very concerned with getting information of the location of the moving robot because using only AprilTags would not be sufficient to know the robot’s coordinates. Therefore, I did research more on Issac ROS and the usage of Intel Realsense cameras to get a sense of a solution.
Here are some links that I’ve found very interesting and useful:
https://www.youtube.com/watch?v=mFLZkdH1yLE
https://www.youtube.com/watch?v=mNTi1osPSIA
https://github.com/ababilinski/realsense-apriltags-calibration-unity
https://www.youtube.com/watch?v=4JuRiPbitdk
I believe using VSLAM on Raspberry Pi is the best solution I could find for locating the robot. I was considering a GPS module, but it has a huge tradeoff with the accuracy, so I gave up on using the GPS.
Finally, just as my other teammates, I have worked on ethics assignment, and we met on Friday to discuss the group portion and submit the answers on Slack.
Regarding the schedule, I am slightly behind the gantt chart because the build was taking longer than we expected. However, I am putting a lot of effort to take majority of my time to finishing up the build so that we can integrate and test sooner. The first priority among all of my classes will be the build.
Therefore, my action items will be:
- Put the acrylic boards on the robot
- Connect wheels to the base
- Work on codes for path finding algorithm
- Start implementing VSLAM
0 Comments