RoboSub 2024 Strategy & Goals
This is a block of text. Double-click this text to edit it.Similar to last year, ARVP will be deploying Arctos for RoboSub 2024. However, the team’s goals now include attempts at all competition tasks - adding the dropper tasks, octagon task, and pinger usage to our scope. To achieve these newfound objectives, ARVP will spend the majority of the year making improvements to our sensor fusion and torpedo system, then add a new claw design and pinger system. Since these task are relatively small, the majority of the team’s focus shall be put towards testing and verification. With hundreds of hours of pool time, ARVP intends to arrive at competition entirely confident of our capabilities.
For the competition run itself, only fifteen minutes to complete all tasks is a challenge requiring both reliability and speed. As such, the following general strategic decisions shall be employed. Firstly, Arctos will rely predominately on machine vision, rather than DVL and IMU data. Our vision pipeline is trained on tens of thousands of images, runs YOLOv8, and is incredibly responsive in updating Arctos’ internal mapping software. Secondly, our mapping system will employ new ‘boundary walls’. These will allow Arctos’ behaviour tree to quickly respond when the vehicle is off course and initiate the proper recovery behaviour to gain additional visual information. Lastly, a new and efficient motion planner has been developed from scratch to allow for faster movement, obstacle avoidance, and more efficient pathing. With these software changes, ARVP is confident Arctos’ navigation will be sufficient to find and action all tasks within the time limit.
For a more detailed explanation of ARVP’s competition goals and task-specific strategy, please see the competition presentation linked below. If you’re interested in the other changes to ARVP and Arctos this year, including structure, verification/validation, maintenance , and operations, please see our technical paper. RoboSub 2024 here we come!