How We’ll Scale the Embark Driver and Put Trucks on the Road

Embark’s new Head of Software JB Passot discusses his experience bringing robots out of the workshop and into a commercial setting, and how Embark will do the same.

At Embark, we have the unprecedented opportunity to solve one of the most complex and promising engineering challenges of our time — building an autonomous vehicle that can reshape our economy and improve safety on our roads. I see this opportunity as a privilege — I’m enthralled by the challenge, and confident that with the right tools and approach, we will get there.

I have the unique experience of having previously brought robots to commercial readiness. During my ten years at Brain Corp, we took an AI research project and turned it into a commercial product. Brain Corp now counts over 15,000 of its autonomous mobile robots operating in public, indoor environments.

But getting to that point required making smart choices throughout the journey. Let’s take a look at how we did it, and what that journey looks like for Embark.

From the Biological Brain to a Robot Brain

My academic background is in neuroscience. My PhD work focused on understanding how different structures of the brain partner to create a stable representation of the environment and help us navigate in space.

Some of this work feeds directly into the same kind of spatial planning that we’re developing for autonomous trucks here at Embark. The way people and the way robots navigate is actually quite similar: we use both receptors that sense the internal state of our own bodies and receptors like our eyes that receive information from outward stimuli.

Early on at Brain Corp, we were a team composed of neuroscientists that focused on building the software for a new type of chipsets that processes the information the same way the brain does.

A few years later, we leveraged that learning to build a new type of operating system designed for robots: we weren’t going to code our robots, we were going to teach our robots. We called it BrainOS.

Here’s a parallel: Imagine you’re teaching a child how to ride a bike. You can describe every little thing it takes to ride, and meticulously explain the mechanics, but the first time the child rides the bike, it will inevitably fall. The best and only way to learn is by actually riding the bike, not by describing the process. That’s the approach we strove to bring to robotics.

Constraining the Problem

In 2015, we decided to bring this novel brain-based technology to automate retail. Over the course of 18 months, we worked with OEM and retail partners and released our first BrainOS powered product, a commercial-ready autonomous software for industrial floor scrubbers.

One of the keys to success in robotics is constraining the problem. At Brain Corp, We selected a narrow but highly-impactful Operational Design Domain (ODD) by choosing to build for the floor cleaning industry before expanding to new verticals. It made the engineering problem simpler, while giving us a focused business case with a clear customer base.

We’re at a similar point at Embark, where we are nearing commercialization of our technology. Self-driving in an unconstrained environment won’t be solved any time soon. But by constraining the ODD to a relatively simpler problem like highways, we have a much more clear path to developing a production-ready product.

Scaling to Commercialization

There are three other parallels between my past experience and Embark that are particularly relevant to scaling a commercial SaaS robotics product:

• We’re not building hardware: There are companies that have been building trucks for decades, and it’s what they do best. At Brain Corp, we saw the same was true for industrial floor scrubbers. Working with existing OEMs was a “no-brainer”. Tech companies aren’t going to revolutionize trucks (or cleaning tools), and it’s far more advantageous to pursue an asset-light model that leverages the best of both worlds — manufacturing and software.
• We’re platform-agnostic: With industrial cleaning machines, we made a clear decision to build a universal autonomy kit that would work across models and OEMs. The Embark Universal Interface embodies this approach in an industry where OEMs have deep ties to trucking fleets and prize those relationships. By building across OEM platforms, and integrating on the manufacturing line, we open the door to a greater customer base and can meet fleets where they’re already at.
• We’re thinking about the supporting infrastructure: Focusing on the autonomy software is just a small part of the puzzle. Where Embark stands out is its thoughtful process to the entire supporting product ecosystem. Once you are ready to scale your product, the infrastructure around autonomy software — things like the manufacturing tools, deployment and maintenance tools, and the fleet and lifecycle management tools — becomes much more important. Embark is already thinking about these pieces and working to deliver a holistic product ecosystem.

Embark’s Culture and Spirit

I strongly believe that when you’re tackling a groundbreaking project like ours, you should be doing it at an organization that has an amazing culture. That’s one of the main reasons I came to Embark. I want to solve this cutting-edge engineering challenge, and I want to do it with people who care as much as I do, and people who I can call my friends. In some ways, I think this will be the game changer for getting to commercially-viable autonomous trucks.

Embark has all the keys for success to commercialize autonomous trucks. I couldn’t be more excited to join a company as innovative and forward-thinking as Embark, and I look forward to moving our engineering organization forward.