Modbot: Interview with Adam Ellison

0
606

Modbot, is a company that wants to provide a complete robot development platform to empower all creators from large manufacturers to individual makers and hobbyists. After witnessing the difficulties of prototyping with the current state of robotics, Modbot decided to create a modular robotics system that would be simple, affordable and agile. With the Modbot technology, they hope to see solutions that will solve real world problems in manufacturing, education, engineering, and consumer industries. We got the chance to interview Adam Ellison, Co-founder of Modbot, here is what he had to say:

Tell us a little bit about Modbot.

We are a San Francisco based startup doing exactly what you might guess: building modular robots. The company was created with one mission: to make robotics drastically accessible. We believe that the market for modular robotics will explode as soon as the complexity of implementing a new robotics system and the costs of implementing a robotics system decrease (the modularity aspect is the key ingredient to lowering these costs). Currently, we are focusing on industrial applications. Industrial and manufacturing customers are already aware of their needs, which allows us to create tailor-made solutions for them while continuing to perfect our product design and functionality. Working with these large firms also sets a very high threshold for quality, which is our priority. Down the roadmap, we plan to expand into small business applications, and eventually into consumer applications as well.

Tell us about your journey from start up to where you are now.

Daniel and I met originally in engineering school at college, and kept in touch over a decade or so post-graduation. We contacted each other regularly, and realised that in both of our careers, we encountered the same obstacles, though our specialties were different. I was looking at the role of automation in the food industry. Daniel was working in defence, building autonomous vehicles. We both experienced challenges in integrating and commissioning actuators and intelligent motion. This common problem is what kicked off the initial conversation that would lead to Modbot. Together, we brainstormed about a solution to this very problem, and of course, it would be something like legos, where you could snap together blocks that represent various functionalities, but at a commercial and robust quality rather than as a toy. There was nothing available on the market, so we set out to build Modbot. That was the conception stage. Together, we formed the company, deciding to jump headfirst into the startup life. In the fall of 2013, we applied and were accepted into a competition called ‘Hardware Battlefield’, which is a Techcrunch event at CES. So in January 2014, we pitched on stage with nine other hardware companies, looking to secure funding and kickstart our crowdfunding campaign. This was very exciting as we were only 2 months old as a company. Although we did not win, one of the judges, Brady Forrest, encouraged us to apply for his hardware accelerator program Highway1, which is based in San Francisco. We were based in Australia at that time, and so declined. However, we came to an agreement and moved the company to the United States in March 2014. That was the start of our existence here. After three, strenuous months at Highway1, we built our very first prototype, the red robot as you might have seen on some of the YouTube videos. We call it the Alpha. It’s our first robot that we demoed at public shows and used to raise capital. Subsequently, in 2015, we started to commercialise the Modbot product.

What were the hurdles you have had to overcome and what does the road ahead look like for you?

There were plenty of hurdles. We are building a very complex system, so the biggest hurdle is getting everything to work at once. There are multiple layers of hardware that need to work together seamlessly, in addition to multiple layers of software combined with communication protocols. All are required to work simultaneously to provide the best experience. Generally, reliability is not a feature that startups are known for simply because companies working on a small scale don’t have the breadth for “what if” cases. I think one of the most difficult steps of our journey so far has been changing from the rapid development mindset, to a quality focused mindset, and then moving back and forth between them. You want to be innovative, but you also want to create a high quality product. Therefore, alternating between these two engineering paradigms has been one of the greatest challenges we’ve faced. Additionally, it takes time to recruit personnel that are the right fit for the company as well as build the logistics side of our business (this is highly underestimated when you’re thinking about growing a startup. It takes a lot of time).

The road ahead is an exciting one. We are on the cusp of our first commercial releases. We’ve conducted a few limited pilots with reasonable results, and are very excited about scaling up. We are in the middle of pilots that involve twenty and thirty units. These experiments are focused in a production environment, and add value to our data and product design process. Additionally, we’re developing a wide range of applications from deburring and sanding, to pick and place and electronics assembly. All of these applications involve the same modules and the same architecture of the product, which is our vision coming true. We’re all for flexibility, which is the centre of our thesis.

How are you planning to bring robots to the masses?

It’s a difficult problem to solve. The plan is to start by integrating Modbot into the ecosystem of a traditional manufacturing firm. The manufacturing process has been around for decades, and we are using this opportunity to bring our product up to industry standards and start selling it to be integrated into these industrial environments.

Our business model and our product architecture greatly differs from a traditional robot developer. Typically, a robot sale is a third of the total cost of the automation process for a company. The integrator would take a third of the value, the robot manufacturer would take a third of the value, and the software developer will take a third of the value. What we are trying to do is to make it simple for the robotics integrator and the customer to do the physical integration themselves. Additionally, we want to make the programming of a robot a simple “drag and drop” type of experience. These two changes will bring down the costs of implementing robots significantly. Over time, we plan to decrease that cost until we reach the point where a small business can contemplate commissioning robots, which can drastically change their economies of scale (previously that was not possible).

The final stage of our plan is to evolve Modbot into a development platform for new applications. For example, in the service industry, when someone wants to build an automated restaurant, they need a platform to develop it from. They would be able to use Modbot components and Modbot software to develop and deploy the necessary type and amount of robots.

What about ROS, is that a competitor to you?

Our mission is to make Robotics massively accessible. We simply want to accelerate that trend. We are using ROS to some extent internally. The most important thing is that we are making our software ROS compatible, so if you have a software stack built around ROS, you can plug in Modbot hardware with a driver that we’ll supply, and it will work seamlessly. There will also be a way to use ROS diagnostic tools and we will create a ROS node that is the entire Modbot robotic system. We require our internal architecture to be extremely robust, so we’ve developed our own. Also, we’ve developed attachment points, where you can plug into the ROS ecosystem at any level of the stack.

What applications do you see MODBOT catering for?

We have already investigated and generated demos for deburring, which eliminates the sharp edges on pieces of metal in a production environment, sanding and cleaning applications, as well as pick and place applications for electronics assembly. These are all existing opportunities, however future areas we would like to explore include the service industry, such as automation within restaurants, the medical industry, specifically with assistive devices and mobility devices, and the gaming industry. We would also be interested in experimenting with the physical portion of VR. With VR, there is an immersive visual and audio environment, but the physical environment is limited to the room that you are standing in. If you add a robot into that mix, you can emulate the whole physical world that’s happening in the virtual environment. For example, if you had a robot under each foot, and one in each hand, you could emulate running, ski-ing, jumping, bycicle riding, holding a steering wheel etc. To me, that’s the next generation of gaming environments, which is not possible because of the costs right now.

What would you say to new start ups that are entering the field?

Do not underestimate the difficulty of product development. Getting the first prototype working is just two percent of the product development journey. Understanding that there are still more hurdles ahead helps. In our minds, when we started Modbot, the first working prototype felt like we were halfway through product development. That was not the case. There are so many challenges to overcome before the product is actually useful to a third party, as opposed to the person who developed it.

The other thing I would recommend is not to underestimate the business side of a start up. The business side is well trodden. There is plenty of material out there to learn from, but if you haven’t built it from scratch before, it’s easy to underestimate the setup, organisation, and planning it takes to create a fully functioning system.

We are excited about the future of Modbot. We hope that people who are contemplating new robotics applications look at what we are doing and see if Modbot could solve some of their software and/or hardware problems that have yet to be overcome.

Robotics News, December 2