That Didn’t Work

In the year since we envisioned Red Triangle Technology we’ve learned a lot. To name a few of our findings:

• There is a lot of technological advancement that needs to happen in the commodity UAV space before certain applications are going to be practical. For starters, battery life on modern aerial drones would cause a substantial drone fleet to maintain coverage over coastal areas and introduce complexity for the operators that have to manage the fleet. Other technological issues exist rooted in UAV hardware itself that limit how much data can be transmitted and at what rate. Lastly, for commodity UAV platforms that are programmable, developers face a pretty serious vendor lock in problem.
• There are a number of legal advancements that need to happen to make UAV applications legal. Today, the FAA has crippling regulations in place that make using UAV technology illegal for most practical purposes. For the most part, a human needs to be piloting a drone, and even then, the pilot still needs to consider no fly zones and human safety. Many coastal areas in California are no fly zones.
• It is tough to navigate city bureaucracies that control funding for state parks and beaches. And even when you do find someone to talk to, it becomes clear the budget is tiny, likely microscopic. Even finding peoples’ times to share domain knowledge is tough. People are busy, we suppose, so that makes sense.

The gist of our findings is that commodity UAV technology has a way to go for applications like our own, and that talking to people about UAV technology elicits yawns and gets eyeballs rolling pretty quickly.

During this same year we’ve also observed some key things.

• We’ve built a scalable, flexible, and easy-to-use sensor processing pipeline that lends itself to machine learning applications, especially applications that make use of video data in convolution neural networks. There is an unfathomable number of applications of this system, many of which still move us in the direction of our mission.
• There are existing sensor infrastructures, e.g., National Oceanic and Atmospheric Administration (NOAA) data sets and Surfline’s syndication service. Leveraging existing sensor infrastructures solves the problem that we were trying to solve with UAV technology, albeit in a different way. Sure, you don’t get the same type of overhead aerial data, but you get something that can be plugged into the sensor processing pipeline to provide value.

With our findings and observations in mind, we’re going to pivot and refocus our efforts. We’re culling our efforts related to UAV until that time when commodity UAV hardware is cheap and efficient, and the legal space is better suited to support new technology use cases. We’re doubling down on our efforts related to the sensor processing pipeline, and re-envisioning products that we can provide with existing sensor infrastructures, such as NOAA’s and Surfline’s. Finding other cost effective sensor infrastructures continues to be a goal.