USER TESTING
During the testing phase of the project, participants found it relatively easy to understand how to control it, primarily thanks to the instructions menu provided on the introduction screen and the clear grid divisions on the control page that explicitly indicated the directions. However, one aspect that caused confusion for many was determining which hand controlled the left movement and which controlled the right movement. This confusion arose because the camera captures a laterally inverted image, which is then processed by the computer. To address this, explaining the control scheme after the initial confusion helped participants grasp the concept of how the handpose.js library maps the tracked hand movements to the car’s controls.
The most successful aspect of my project is undoubtedly the mapping of hand movements to the car’s controls. The serial connection between the Arduino and p5.js worked flawlessly, ensuring smooth communication. However, there is room for improvement in the implementation of the hand tracking model. Optimizing it is necessary to ensure that the regions of the canvas corresponding to the various controls are accurately defined. Also, i realized that the lightning conditions also greatly affects how easily the model is able to track the hands. In an attempt to improve that i will either go back to using posenet because i realized it is more accurate as compared to the handpose.js I am currently using.