4 Must Know SUAS Challenges
February 18th 2023
Didn’t think we were gone, did ya? It’s been a restful winter break since we last released an update, but now Ascend is back, and we will be working hard as ever.
Truth be told, being an NTNU student and a part-time drone developer can be challenging. Hence at Ascend, we take a break for exam season, then kick off the new year with a work week to welcome everyone back and to start building our next drone. That said, our new drone is a bunch of unboxed components and some blueprints right now, which isn’t too exciting. So instead, we would like you to know why we’re building it and what’s the deal with the SUAS competition.
What is SUAS?
SUAS refers to Student Unmanned Aerial Systems, the largest global competition for student teams to make a functioning drone that follows industry trends. Every year, the Maryland Seafarer Chapter releases new challenges! These challenges require contestants to consider all aspects of drone engineering: Technical design, software integrations, and real-life demonstration of a custom-made aerial robotic system.
Unlike other competitions, this is the first time the Marketing team gets to shine with collecting and writing technical specifications. As per entry requirements, all contestants must prepare a 15-20 minute video presentation to enter the contest. Once the presentation is approved in early May, we will compete on-site with nearly 70 international teams in Maryland, USA.
Challenge 1: Autonomous Flight
Autonomous navigation is the most fundamental challenge for this competition.
To meet the standard accuracy, the drone should aim for more autonomous flight lengths and designated waypoints. All flights must maintain a minimum height of 75 feet/23.86 meters above each waypoint. In other scenarios, flying drones will perform various actions (e.g., air delivery) for up to 10 miles/16 kilometers across all waypoints. Long-distance navigation and constant height control require solid and stable remote sensors and data/signal transmission. That’s why cross-team collaboration between Autonomy, Control, Perception, and DevOps is crucial to ensure navigation accuracy. As a rule of thumb, contestants must implement a decent amount of artificial intelligence for the drone to make smart decisions in path-finding scenarios.
Challenge 2: Obstacle Avoidance
Flying around with nearly 70 different drones and mission props could be chaotic. During autonomous flights, every contesting drone must autonomously avoid colliding with each other in the shared airspace.
The avoidance feature is often mandatory for smaller drones since they will fly low and encounter more obstacles. But for SUAS scenarios, each drone must find ways to avoid flying aircraft when multiple drones are dropping items at the same waypoint. Therefore, each contestant must come up with innovative solutions for such cases.
Challenge 3: Object Detection, Classification, Localization
The following scenario aims to demonstrate the drone's computer vision capability. Within the range of fly zones, there will be several standard-shaped objects, such as squares, circles, and triangles, in different colors.
The onboard computer must be able to identify these objects and classify them into several categories, ranging from colors to standard or emergent objects. The entire process is conducted within the built-in computer, but what’s the winning goal? While the objects are being detected and classified, the drone’s software should collect the exact locations of these objects. Then, we must fly the drone to record and generate a complete virtual map.
In addition, the drone faces the challenge of detecting and recognizing objects while flying rapidly. This problem is where Perception brainstorms a solution for image accuracy. For example, would a well-trained neural network identify a high-speed moving object through blurred imagery? Or would an expensive camera or sensor compensate for the technical difficulties?
Challenge 4: Air Delivery
Then comes the most challenging part of the competition: the airdrop delivery feature. This is deceptively hard as it is a combined operation of all previous challenges. A stable airdrop system requires the best performance on the precision of our control systems, the effectiveness of our collision avoidance algorithms, and the reliability of our mechanical designs. The goal is to deliver and drop marked payload at 5 designated waypoints within the flight zone.
Alternatively, each team can also load and drop all payloads in one trip. Teams will receive more scores if the drone can deliver all payload objects faster. The heavy-duty route could reduce unpredictable errors in longer runs. Though it will be risky if the drone fails to deliver the full payload and loses the entire scoreboard!
Furthermore, there’s more to what makes this task particularly challenging. The essence of the airdrop event is, in fact, a test to see how well different systems support each other. For instance, contestants will only receive a description of an object placed at each waypoint. The drone must first identify the precise airdrop location through the onboard object detection system. After localizing the coordinates, multiple contesting drones will simultaneously drop payloads nearby. This is where obstacle avoidance does the trick to ensure the drone’s safety and stable air delivery.
Overall, the SUAS competition closely ties to real-world needs. From autonomous navigation to obstacle avoidance to object detection and airdrop. Every challenge is a great way to learn by thinking about and doing engineering the way it is done in the real world. It is also an excellent experience for dynamic cross-team collaboration.
The upcoming drone for the SUAS competition is one of our most ambitious projects. We can’t wait to reveal this bad boy. The stakes are high, but we are working hard and are confident we will succeed. As we launch our first workshop week in early January, we continue to ascend above and beyond!