Billions of dollars have been spent by a variety of tech and automotive companies over the last five years, trying to solve the issue of self-driving.
Tesla’s Elon Musk has been promising it ‘next month’ for several years now, governments are skitchy on approving existing software and tech because people keep getting run over by it, and half the drivers on the road want nothing more than to be driven to work while they play Candy Crush in the back seat.
So what’s taking so long?
One problem.. recognition of threats.
Anyone who has driven in a Tesla has experienced the drivers’ screen, showing hazards where they aren’t, or the auto brakes kicking on unexplainedly, or the car suddenly deciding that a lane change would be great right now, whether you like it or not. It’s ‘imperfect’.
In an earnings call earlier this year, Musk dismissed LIDAR laser systems, radar, and all manner of other tech his competitors are using to detect hazards in self driving, saying, “It is obvious that our solution with a relatively low-cost inference computer and standard cameras can achieve self-driving,” Musk said. “No lidars, no radars, ultrasonic. Nothing.”
Earlier, he’d dismissed LIDAR as “a crutch.”
Yet on that same earnings call, Tesla revealed it had bought so much LIDAR the previous quarter that they were responsible for a full 10% of their supplier’s revenue. Ongoing performance issues are no secret.
Waymo’s solution is to flood the field of vision with overlapping tech. They claim their sixth-gen robotaxi will feature a suite of “16 cameras, 5 lidar, 6 radar, and an array of external audio receivers (EARs).” That’s a lot of tech, and doesn’t even include GPS, which is notoriously inaccurate for anything inside 15 metres.
Lasers, radar, GPS, satellites, image recognition, even the internet itself – relying on all that to make things happen is like using an Apache attack helicopter to open a jar of pickles.
What everyone actually needs is relatively simple and a lot more analog – we need intelligence that can identify objects in the distance, figure out what the objects are, and use simple math to understand how far away they are.
SPARC AI (SPAI.C) has the tech, and has been quietly proving it out in settings non-military and otherwise.
“The company is pleased to report that it has successfully deployed the Sparc AI algorithms on the Parrot ANAFI USA drone and successfully completed a series of autonomous flights that demonstrates the drone self-navigating and flying autonomously without GPS, lidar (light detection and ranging), radar, satellite or image recognition software. Setting up the autonomous flight is done whilst the drone is deployed in the air and activated within seconds of identifying the target location. There is no need for complex flight path orchestration software or building the flight path offline and then uploading to the drone.”
Sparc has taken the route far less complicated in developing their tech, putting it directly on an artificial intelligence chip that can be put into a cellphone, or a high end drone, and has specifically been used in the Parrot ANAFI USA drone thus far, which is a popular off-the-shelf military/first responder model already in wide use.
The [$14,000] Parrot ANAFI USA GOV drone is designed for first responders and offers the same high-end security, durability and capabilities as Parrot’s short-range reconnaissance (SRR) drone designed for the U.S. Army. The drone is manufactured in the United States and trusted globally by defence and security organizations.
Instead of building a ground-up drone themselves, or hinging on one model and hoping end users like it, SPARC’s AI methodology was to just build the microchip and allow it to piggyback on any drones already out there, keeping the end user price down and hardware flexibility high.
Once installed and flying, the operator can identify an object and set the drone to go get it without having to train it, or code it, or hand-hold it, which is outstanding for military uses.
I’ve represented companies in the military drone space before, and their big problem is never in identifying the threat, it’s staying stealthy once that threat is identified.
If those products use radar, they give off a signal that betrays their location. Laser likewise. GPS can be jammed. Wifi same. Satellites dont like cloud cover.
But SPARC’s tech is hamstrung by none of those potential tripwires. It just looks, identifies, assesses, and spots threats, and then heads for them as required.
If the target moves, it’ll follow. If the enemy jams the airwaves, it’ll keep going. If the enemy looks for a ‘ping’ from the drone, it won’t find one. This is rad stuff.
SO HOW DOES THIS APPLY TO DRIVING?
It doesn’t, yet. The tech is at an early (working) stage, and needs to be tested and improved in more environments. Thankfully, that’s occuring.
Having reached this critical milestone, the company is planning to extend the technology for several use cases. For instance, using algorithms that provides instructions to the Sparc AI-enabled drone to self-navigate based on its environment for defence, border control, search and rescue, and security.
Military uses are great, because the military has a massive budget and pays on time.
First responder uses are good, because they have a lot of use for spotting objects in the distance, and they pay on time.
Self-driving though – that’s the whale. If one Tesla or Waymo or BYD or anyone else spots SPARC’s tech and realizes it solves a lot of their problems, SPARC could find itself in a licensing war where it wins a small piece of every product purchase out there in a quickly growing industry.
In short, it might cost the robo-taxi crew more NOT to license the SPARC tech than it would to play ball.
And they’re not wasting any time:
Sparc AI’s target acquisition system is a covert and highly flexible software installed on any camera device for drones and large surveillance systems. Used for making tactical and mission-critical decisions, the technology can determine the location and distance to any distant object, without connection to GPS.
Sparc AI’s autonomous flight solution allows customers to deploy and fly drones autonomously using Sparc AI algorithms in GPS denied environments. The company is also working on extending the capability for autonomous flight without a controller and connection to base.
SPARC’s newsflow over the last year is clean and tidy – all technology advances and delivered upon promises. A small amount of financing, and a share price that’s doubled in that time.
Here’s the kicker: It’s only a $5m market cap, even after the double.
SPARC hasn’t blown itself up on the hype train of AI and hasn’t gone bonkers trying to attach itself to ticker symbols it hasn’t earned. It’s just a really nice little tech development story which has kept costs low, moved things along, and finds itself smack dab in the middle of one of the fastest growing tech needs of the last fifty years.
At that price, you can be a bully at the table, and any decent sized piece of news could move the stock significantly.
Great time to put it on your watchlist.
— Chris Parry
FULL DISCLOSURE: SPARC AI is an Equity.Guru marketing client, and we may buy stock on the open market
