We continue with Ben and Mike’s discussion about the intersection of the armed forces and agriculture. The original audio version of the podcast is available at the end of the article, but here’s the second of three installments of its transcription (and the first if you missed it):
Mike DeSa: Why do you think unmanned systems and cyber are so high on the desirability of things when it comes to professions with the Israeli military?
Ben Alfi: On the unmanned systems, first of all, there are no funerals. Pure and simple. And, that was the beginning. Then, it was also an understanding that if you want self-independence as a country, you need to have self-capabilities. And making an aircraft, just like the decision to be an OEM: do you want to build a tractor, or do you want to build an autonomous vehicle? So you don’t need to deal with the hardware and will never be as good with the hardware, but creating the unmanned system that’s good enough to operate and have the variety to operate was a 20-year program with strategic capabilities. It’s a huge revolution, amazing success how we implemented those capabilities and numbers. When I started my job, a low percentage of it was unmanned systems. Today, more than 80% of the flight hours are unmanned systems, and we’re flying four times the flight time. You can create much more from it. It was the correct way to do it and from a budget and R&D perspective, it was the place to invest. And the second was cyber, cyber was the understanding that today everyone understands the lethality or capability of cyber warfare whether for defense or offense, but, again, it was a decision from 20-25 years ago. This is why you see the effects also in the civilian markets on cyber security and unmanned systems.
MD: Can you talk a little bit about what some of those platforms, the sensor capabilities are or some of those unmanned systems today and the kind of data they can collect?
BA: Everything is open source, but the idea is that we understand that I’m saying things that today everyone understands. We talk about multi-sensors, not relying on one sensor, the ability to work on the whole spectrum of visual and RF to get situational awareness of what is happening in an area. You have a lot of Tier 2 companies in Israel that are creating those sensors, and some places you can buy from abroad, but most of the sensors will be developed by Israeli companies. The whole world of composite materials was crucial to creating unmanned systems and big bodies that can take a lot of payloads. These are the areas that we see huge capabilities. Israel and the US are the biggest exporters of unmanned systems to the western world.
MD: Was it this work that you did in the service on unmanned systems that ultimately lead you to the creation of Blue White Robotics and agriculture?
BA: When I retired, I had an amazing chance at the age of 42 to have a second life. And my decision was that I wanted to create a worldwide impact from Israel in the civilian market. Meaning that I was not going to work for a US company or abroad, but I was going to do something big from Israel that would affect the world, and I didn’t want to do anything connected to the military. I had enough of that, and I think we can do other things. When I evaluated what was the toolkit that I was coming with, what do I have, I understood that we are on the verge of the revolution that we had in the late 90s up until now. It’s starting now in the civilian market. For sure, I understood that I needed to create a company dealing with how to be part of the revolution of unmanned systems in any industry. In the beginning, we were really open. “Let’s look at urban mobility. Let’s look at agriculture. Let’s look at oil and gas. Mining. First responders.” The first two years, the door was open. We looked and checked on everything. So this is where I arrived. I understood that what we had done, we could be accurate and implement things in the civilian world. And what we also understood: not many people know how this story ends. We know how this story ends. There is a way to correctly get to the end of that story. We made so many mistakes at the beginning, and I see the same thing in other places that we’re trying to implement autonomous capabilities. Without having a good way of implementing these capabilities, it’s not going to happen.
MD: Where have you seen those other industries fail in the implementation of autonomous technology and why?
BA: First of all, if we take ad bits from 2017 where everyone’s talking about urban mobility, autonomous vehicles running around, everything is going to be unmanned. And, I came to the civilian market and asked, “Where am I wrong here? Where am I wrong?” The environment is so complex. It’s not like flying in the air. Making it real-time for computers is so complex. There is a big difference if you’re in an area that is unmanned or an area that is unmanned and manned together. How can I know that this child on my right-hand side won’t just pop into the street? It’s not going to happen until 2030 at least. Another thing that I’m seeing is that there will always be some type of monitoring station that is monitoring all of those vehicles, and you need to have someone decision making. This is what we understood from the beginning. This is what we have developed.
MD: So you’re ultimate decision to go into agriculture as opposed to any of the other industries that you described was, correct me if this isn’t right, was a factor of how quickly you could implement something, the impact it could have. Were those the primary drivers as to why you settle on agriculture and not all of those different areas that you just described?
BA: In agriculture, all the stars aligned amazingly. First of all, the pain. The labor shortage is very high. The precision of the labor: good tractor drivers. You want them to drive at a certain speed. It’s not easy to drive at that speed, and if you’re not driving at that speed you need to spray again, or you need to do another action again. And, the only reason in high-value crops where you spray so many times a year, the only reason that we have people near those materials is because there is no other way to do it. But, there’s no reason to do it like that if we have robots. The pain is very high on that part. So that’s one reason. Second, all the missions, and I’m talking military-style, are predefined. You know where you’re going. You know what you need to do. The only decision that you need to make is to abort the mission or there was some kind of malfunction. There is no decision-making that’s needed in that regard. Third, you need a huge amount of vehicles in a certain area. You have a farm that’s two or three thousand acres of vineyards or citrus, and every hundred acres has a vehicle, and you drive in four shifts. They can be four shifts of John Deere, New Holland, and Kubota mixed together – a mixed fleet – doing all those missions. And, you need to have the ability to either mow, spray herbicides and pesticides, discing. The need for a solution and ability of what we’re providing told us to go all-in on agriculture.
MD: Describe that solution in more detail for us.
BA: We have three pillars that our solution is based on. First, there’s a software platform that enables you to operate different types of vehicles in the area of agriculture. You can operate different types of tractors and other vehicles in the area. Command them, tell them where to go, and while they’re collecting any data that is needed, we can distribute the data to whoever wants it. The second thing is that we take every tractor and implement an aftermarket kit on the tractor. We change this vehicle from manned to either manned or unmanned. You can still drive it manually with the flip of a switch, but you can also drive unmanned. We connect this vehicle to our platform, and then you can operate all of those multi-missions that you want to accomplish on your farm. The third thing is that let’s say we have a grower, we took his vehicles and changed them to autonomous, and it’s all connected. “How do I learn and create adoption?” The adoption, you can’t throw those technologies on the end-user and say “good luck.” The adoption is crucial, and we’re there to bring it as a service from Day One. In the beginning, when the tractors operate autonomously, we are the operators, and we do on-job training at the same time.
MD: When you line it up on a row, you essentially, hit “Go.” Is there somebody back in this command system that you talked about controlling it, or is it using Lidar or Radar to determine where the left and right it can go? How does it know when to turn around at the other end?
BA: There comes a lot of things from our offroad background: we don’t rely on the man in the loop for the mission execution – only for decision making and safety when something occurs. You cannot rely on having communication all the time in these rural areas. So you will have several types of sensors, whether that’s a visual sensor, lidar, GPS, etc. that will be part of the decision-making for who is voting or understands where you drive. Driving in a row, GPS will not help you over there when you have just a few inches to drive in. So we need better accuracy from the other sensors that give us this understanding of how to go and operate. And some other sensors will take the job of performing redundancy. In unmanned systems, redundancy is always the issue. And when the vehicle says, “I don’t understand, I have a problem.” Then, it will stop, we open a video, and we can go in and tell it it’s not a tree trunk, it’s okay, you can keep on going. You hit Go and keep ongoing. We’re not doing teleoperation, we’re doing telemonitoring.
MD: No, that’s an important distinction because I was wondering how you deal with communication infrastructures that are not advanced or get interrupted by satellites or you might only have 2G capability instead of the required 4- or 5-. But, it doesn’t sound like you’re needing a communication architecture to transmit data back for any other kind of system functionality. It’s using onboard systems to be able to execute its mission until it gets to a point where there’s a safety issue or it recognizes something may be a problem, stops, and then uses some kind of outreach method to be able to touch back to an operator somewhere else that can confirm or deny what the issue is and figure out what the next steps may be. Am I understanding all of that correctly?
BA: For the single vehicle, communication is not something that is needed for operating safely, but when you want to do fleet management and coordinate and do an orchestra of vehicles running in the same area, communication is crucial.
MD: Talk about that. How do you enable that?
BA: We came from Israel. In Israel, any rural area is near a town because everything is small. And we said we would do it [here] like we do it in Israel – using cellular connectivity and everything is good. And then we came to a new reality in California. Okay, it’s California. For sure there is communication but also no. We had a problem and had to do a quick solution in that area. What we have done, we are able to either work over the cloud through cellular and wifi, or we go through RF. We created RF stations to create a local server that you are operating.
MD: Okay
BA: What we have done in the last year was a huge effort with T-Mobile, Intel, and the 5G OI Lab is creating local CBRS networks. You create your own local cellular network where you don’t have cellular reception.
MD: It’s almost like the creation of a mesh network.
BA: Definitely, yes. The mesh network that we did at the beginning with RF has a lot of limitations because it doesn’t speak the same language as a cellular network. So it’s a second-degree mesh network on a cellular language. And then moving from local servers to a cloud-based server is easy and transparent to the operator. And, it’s transparent to the vehicle so you don’t need to do two research and development programs. You use the one solution that is suitable and flexible.