top of page
Search
  • Adwith Malpe

Conclusion to My Capstone Project

After 6 months of hard work and numerous coffee stops, I am proud to announce that my team and I have successfully completed building the Telemedicine Robot. This semester was filled with challenges that greatly tested my flexibility and different skillsets in developing the autonomous robot while working with clashing mindsets. Some of the challenges faced during this semester include accessing the DeepRacer's OS and training the model, printing the 3D modeled trailer within a time constraint, and implementing functionality into the Telemedicine Robot's mobile application.


Training the Reinforcement Learning model took a significant amount of time as I had to determine the best hyperparameter configuration that would improve the neural network's learning capabilities and increase its response time. The cost of training each model increased every time I created a new model to be trained on the DeepRacer console. The challenge here was to train the model long enough while remaining within the project's budget. After spending a significant amount of time reading and understanding the responsibilities of each hyperparameter on documentation provided by AWS while also looking at past projects, I was able to identify the best hyperparameter configuration that improved the reaction time of the DeepRacer, which successfully functioned during physical training. Accessing the DeepRacer's OS proved to be a challenge as I initially did not have the equipment required to connect to the vehicle. After some resourceful thinking, I realized that I could use the TV as an external monitor and connect it to the DeepRacer with an HDMI cable that I purchased. This idea helped me as I was able to open up Ubuntu (DeepRacer's OS) on my TV screen and update the software manually, which allowed me to control the vehicle properly and upload any trained RL models.


The next challenge, printing the 3D model within a time constraint, was overcome solely through communication. I was able to work with my Capstone Professor who provided me with a contact to the CIDSE 3D Printing Labs. I was able to work with my teammate and reach out to this contact, who readily accepted our CAD file and informed us that our trailer would be printed within a week since the material composite structure we chose to print in was plastic (plastic usually can be printed within a short amount of time). One week later, we were able to pick up the printed trailer that was broken up into components and use it during the physical testing process.


After this challenge was overcome, I allocated more time into working with another teammate in designing the user interface for the mobile app. During this phase, I drew out a basic prototype design that displayed all the commands and features a user should be able to interact with in order to control the robot. The challenge with this task was implementing all functionalities into the app. Due to only two sprints remaining in the semester, my teammate and I decided to develop a general wireframe that showcased what the intended functionalities are and how a user will interact with the app.


All in all, my team and I successfully developed and merged all components into an integrated system that can be used to transport medical supplies to different rooms within the ICU of a hospital. If you have any questions about my experience, feel free to email me anytime! In the mean time, here is a video of everything I accomplished with my team.



42 views0 comments

Recent Posts

See All

Comments


Post: Blog2_Post
bottom of page