For more than a year now, the coronavirus has changed our daily lives. In HoverGames Challenge 2, “Help Drones, Help Others During Pandemics”, we asked contestants to come up with innovative ways for drones to help people in need, improve crisis management or even directly prevent the spread of disease. We are thrilled to tell you that our talented contestants have put their creativity to the test and have emerged with compelling solutions for drones to help, from maintaining quarantine zones and protecting crops, to helping people mentally cope with the perils of isolation.
Whether it’s flying entertainers, vaccine delivery or drones that give farmers the tools to protect crops against wild animals, we are excited to see the qualified and innovative ideas in HoverGames Challenge 2. These ideas show that even during a crisis, human spirit prevails, and that compassion, solidarity, perseverance and problem-solving skills of engineering scientists can lead to innovative and life-saving solutions.
Check out the highlights of HoverGames 2 in this video!
In HoverGames Challenge 2 (HG2) participants were asked to build and test their idea using the complete NXP robotic drone kit and incorporate the experimental “NavQ” vision system and mission computer based on NXP i.MX 8M Mini. It was asked that they share the code so that their ideas can also inspire and literally "help others".
The Winning Entries
Food shortages have been a real threat during the pandemic. Many of us remember empty shelves of pasta, flour, yeast and even powdered sugar. Due to closed borders, harvest workers were initially unable to enter many countries. Farmers were missing the workforce to harvest and manage their crops.
In addition to wild animals potentially spreading disease to crops, it was suspected that the coronavirus itself was transmitted to humans through the consumption of wild animals. Our first-place winner, Team SCAREcrow, looked at how drones can be used to autonomously protect crops and livestock from wildlife, and prevent wild animals from transmitting infection or disease to crops and livestock.
SCAREcrow consists of modular and scalable software parts: agents, controllers and base station. In this scenario, the agent is the drone with the FMUK66 PX4 flight controller and the NavQ mission computer. The controller is a ground-based artificial intelligence (AI) process that is analyzing and working with the image data being provided by the agent.
Meanwhile the base station software runs coordination between the controllers and the agents and ensures the agent and controller are subscribed to each other for this short-term activity. When not in use, agents (drones) and controllers (AI) return to a subscription pool.
More simply put, when wildlife is detected in the field, one or more drones are deployed to stream back their video and location information (the drone can also offload some of the Neural net processing). The base station connects them with a controller software AI process that then manages them while they “herd” animals away by triggering the drone(s) to disperse wildlife. Once the animal is no longer a threat, the drone returns home and waits for its next job.
This system, that sits passively in a farmer's field, watches for animal intruders, and sends the NXP drone to scare away potential threats, can be easily scaled for coordinated control of multiple drones by multiple agents to more effectively cover larger keep out zones.
This idea applies neural networks with reinforcement learning and shows a new, innovative use case for drones in agriculture. Currently, most of the drone use for farmers is either related to fertilization of crops, or to specifically monitor crop health. Using drone technology to watch fields for destructive pests opens another new application area. Additionally, this project was clearly written, documented, and the ideas presented in a clear manner. The story is well told and enhanced through a series of YouTube Videos as well as a Wiki website that walks you through the steps needed to recreate this project.
Congratulations, Team SCAREcrow! You Are the Winner of the HoverGames Challenge 2!
Our second-place winner for HG2 is no stranger to HoverGames and was the winner of the HoverGames 1 challenge “Fight Fires with Flyers”. Dobrea highlights that although there are now different vaccines on the market that protect from an infection with Covid-19, it will still take some time before the entire population has been vaccinated. Until then, social distancing, washing hands, wearing masks and keeping rooms ventilated is still considered the most important preventive measure to protect yourself and your loved ones from an infection.
Dobrea’s idea focuses on an autonomous drone to sustain and support quarantine zones. As a starting point, Dobrea built a development system based on the HoverGames drone, the robotic drone Flight Management Unit, and Linux companion computer. By using the NXP FMUK66 running PX4 autopilot, the drone will carry out a pre-programmed autonomous flight around the quarantine zone and return to the landing point. If movement is detected via the intelligent system, the drone will send an update to the base station. The system can also support diverse types of applications such as transportation of goods.
The highlight of Dobrea’s entry is the detail and quality of each step-by-step instruction, accompanied by a very thorough documentation. There were several carefully controlled experiments resulting in improvements to the telemetry system range and GPS noise floor, as well as performance analysis of neural networks running on various processors. Dobrea explains details of using OpenCV and even updating the NavQ software distribution Linux image.
Dobrea shows how to take advantage of the impressive PX4 software ecosystem including communication between FMUK66 and NavQ and Groundstation based on MAVSDK C++ library and the mavsdk_server, and then later by way of custom uORB messages and MAVLink with custom Python code running on a Linux host.
The level of detail included in the submission is impressive, as is the extent of how he “helps others” with his instruction.
During the pandemic, social isolation and resulting existential anxiety seems to be leading to many people suffering from mental exhaustion and anxiety. The added fear of an infection can also prevent many people from going outside for physical activities. This is especially true for those living in cities and crowded places.
Fabio’s drone makes outdoor activities safer by seeing where people are walking and uploading the data to a map application, highlighting how frequented an area is. This allows people to make an informed decision and identify safer and less crowded places to go for a walk.
The drone uses object detection and tracking algorithms using OpenCV to observe walkers and upload the aggregated path travelled to a mapping application. The map shows a 3D 4-meter bubble of the walker that can help others make precise movements when they decide to go out. Since the camera Fabio used turned out to be too heavy for the drone, he 3D printed a base mechanism which can support dampers and handles both NavQ Board and Coral Camera to save space for the fly controller at the center of the drone.
Fabio’s idea was another excellent example of documenting a project well. The included images, source code and instructions for creating and using custom messages also shows the versatility of this software and hardware ecosystem. Carefully constructed vision-based applications of drones can supply useful research data and inform the public while supporting and addressing privacy concerns.
We are thrilled about the high number of quality submissions: it was incredibly challenging to judge the top 3 and a tight race within all top ranked entries. Thank you everyone for their great contributions, online collaborative discussions, awesome videos and for taking part. In addition to the top 3 prizes, we’re excited that we have several other Special Award Winners. Stay tuned for a follow-up blog article announcing these winners!
This is the second NXP HoverGames Challenge “Help Drones Help Others During Pandemics”. Read more about the winners of the first challenge, “Fighting Fires with Flyers,” in our first blog.
Mobile Robotics, Drones and Rovers, Program Lead, System Innovations at NXP Semiconductor
Iain Galloway, P.Eng., holds an Electrical Engineering Degree from University of New Brunswick, Canada, and has more than 25 years hands on experience as an electronics embedded design and field support engineer. You can connect with Iain on Twitter at @iafgalloway.