One of the risks that we are concerned about is the integration of the ROS packages we are using. As we have not used ROS before, we are a bit concerned about how well different ROS packages/ components work together. We are considering running part of the path-planning algorithm on the Jetson Nano Orin and the rest of the motor control code on the Raspberry PI if we used Isaac ROS (a set of hardware packages that provides image-processing pipelines & allows for 3D representation of a robot’s environment). Isaac ROS has been highly optimized for NVIDIA GPUs and NVIDIA Jetson platforms, hence, it would be most ideal to run this on the Jetson Orin; however, we must also consider computational efficiency in this regard (as tiny yolo is also running on the Orin). It was mentioned online (on the isaac-ros page) that Isaac ROS is compatible with existing ROS 2 nodes (mostly what we will be using for motion control), hence, we expect it to run; however, it is something to keep note of.
Another risk that we have been considering recently is the build of the system. We built the frame of the system and have cut out acrylic boards for the other components of the system; however, we are concerned about the timeline. The construction of the robot has taken a bit longer than expected. Building the components and having all of the parts (connectors and acrylic boards) come in time for the construction extended our expected deadline. However, there has been consistent progress on the build and so we expect to have the pick-up mechanism integrated with the frame by the end of next week. That way, we hope to present a complete (physical) component for interim demo.
We did not make any changes to the overall physical design of the system, but in terms of the path planning, we decided to use Isaac ROS and April Tags. This decision might change depending on how we find Isaac ROS and its packages.
0 Comments