This week we prepared for our interim demos this week to show the hardware and software progress for our robot. On the hardware side, we showed our intake mechanism working with a drill. The intake mechanism works with full sized/fixed size water bottles very well. We still need to determine if we can handle variable/crushed sized water bottles. This will be very difficult because we aren’t able to move the intake mechanism up and down unless we add some other motor/spring mechanism, which is very difficult.
On the software side, we were able to put together a prototype software pipeline incorporating inference, distance readings, and object tracking to implement the robot navigation control. Next, we will be working on using the distance readings to move the robot forward until it reaches a distance that it can effectively pick up a bottle. We will also need to work on finishing the motor control aspect of the robot so that we can activate the intake motor when signaled by the Jetson. Inference still takes 20 seconds which is highly unideal so we are also actively trying to reflash the Xavier so that we can get an updated JetPack SDK and use CUDA for GPU acceleration of our ML task. We believe we are slightly behind on schedule for the motor control (motor driver) and navigation + movement algorithm aspects, which we’ll be working on this week.
0 Comments