Drone Roundup: 3 Innovations Take Smaller, Smarter Drones to the Sky

Both industry and academia have set their sights on drone innovation at both the hardware and software levels to leverage air-borne technology in new ways.  

 

Drones such as the Nokia drone are becoming an extremely valuable asset for designers of all disciplines. Image used courtesy of Nokia

 

Drones have long been used to benefit applications requiring birds-eye or through-the-wall vision or applications such as agriculture, where aerial maneuverability can offer a simpler solution that saves time and money. As the number of drones increases, however, sophisticated and robust solutions will be necessary to ensure maximum efficiency and minimum interference.

This article examines three new drone innovations and provides designers with a sense of how smarter drones could enable new applications in the not-too-distant future. 

 

Turnkey Drone Networks

Kicking off the roundup, Nokia has announced the Nokia Drone Networks solution to enable customizable deployments of its “drone-in-a-box” hardware. Already holding a CE certification, the drone network solution comprises an end-to-end solution including drones, docking stations, and edge data processing for data-heavy applications.

 

The Nokia drone and docking station allow designers to focus on new applications without having to design from the ground up. Image used courtesy of Nokia

 

Nokia has targeted several use cases that would benefit from a plug-and-play drone solution, including disaster response, construction, and safety monitoring. The customization of the package allows designers to include only what they need while allowing for versatility with public/private 4G/LTE and 5G networks. The Nokia solution even supports third-party applications via an open API framework.

 

Liquid Neural Network Navigation

In a recently published journal article, MIT researchers demonstrated robust flight navigation using so-called liquid neural networks (LNNs). These networks, while similar to other neural networks in that they “learn” from training data, offer a unique advantage for autonomous drone-based flight. Instead of learning from all data, the liquid networks “distill” the data to focus on the features critical to accomplishing the objective (for example, fly-to-target tasks). 

 

MIT’s liquid neural networks leverage artificial intelligence to extract task-required information without becoming sensitive to noise or environmental parameters. Image used courtesy of Science Robotics

 

While disregarding superfluous data may sound wasteful, it allows the LNN to adapt to unseen environments remarkably well. In addition, the LNN learns continuously, allowing for a theoretically improved performance over time.

So far, the LNN-enabled drones have been tested in controlled and straightforward situations. After demonstrating adequate performance in more realistic and complex environments, MIT envisions that its liquid neural networks could greatly improve autonomous drone capabilities in applications such as package delivery or environment monitoring. 

 

ModalAI Shrink AI Processors

Finishing off the roundup, ModalAI announced its VOXL 2 Mini solution, bringing powerful AI performance to an all-new form factor built for the smallest of drones. AI hardware is generally thought to be big and power-hungry, making AI-enabled drones a unique design challenge. The release of the VOXL 2 Mini gives designers another tool to use for their smart drone solutions.

The VOXL 2 Mini sports similar specs to its predecessor, the VOXL 2. What is notable, however, is its remarkably small area of only 42 mm x 42 mm. This creates a drone-AI solution that is optimized for size, weight, and power without sacrificing pure performance.

 

The VOXL 2 Mini (center) and VOXL ESC (right) offer designers a compact solution for drone-based AI, all within an area similar to a cookie. Image used courtesy of ModalAI

 

The VOXL 2 Mini can be used alongside the VOXL SDK to accomplish objectives such as GPS-less navigation, obstacle avoidance, target recognition, and SLAM. With the addition of the VOXL ESC, the PCB stack can squeeze in even more functionality using the same area. 

 

Smarter and More Capable Drones

While each of these drone developments addresses a different fault in current technology, they are all indicative of commercial drones becoming smaller and smarter, both of which will improve integration moving forward. 

New applications will also emerge following the continued innovation in drone-related technologies. Examples include smart cities or venues, distributed measurement, or autonomous courier services. As is typically the case, these advancements represent new tools that may become available to designers to more easily move technology to the skies.