The increasing demand for drones for inspection, surveying, monitoring, and surveillance applications in the field of agriculture, infrastructure, telecommunications, mining, etc. is propelling the growth of the drone analytics market. Omnipresent Robot Tech is a company working towards developing robots and industrial UAV/drones using very popular solutions like machine learning, virtual reality, computer vision, etc. to provide video analytics, industrial inspections, and defense. The company has also got the approval for e-commerce drone delivery.
In an exclusive interview with Circuit Digest, the founder and CEO of Omnipresent Robot Tech, Mr. Aakash Sinha, shared details about the company, the services they're providing, and much more.
Q. Tell us about your journey with Omnipresent Robot Tech, how did the journey start?
Few years back, I was working at iRobot Corp, which is one of the biggest robotics companies in the world. While I was there, I proposed them as to why we don't expand the iRobots business in India and I did a little stint with them to expand their business in India. At that time, I realized that in India, there are a lot of new technologies like robots, drones, AI but there aren't enough people providing it. Around that time, I felt maybe it's a good time to take a chance of starting my own business over here. We started working with DRDO Labs, and ISRO. We were part of Chandrayaan I and subsequently Chandrayaan II. We incorporated a company over here in India and started developing robotic solutions.
Q. From building drone surveillance systems to writing AI navigation software for the Chandrayaan mission, Omnipresent has shown its expertise in a wide range of fields. How would you sum up the products or services offered by your company?
Companies like Wipro make software for banks, hospitals, government. Similarly, we have been building a variety of robots for a variety of industries and segments. Our strength mostly lies in the software that powers the robots which is the AI. It's this AI part that we build for robots, and it can be used for a variety of robots. On the surface, it might look like we are doing a lot of different things but in actuality; we are essentially building the brain of a robot that could be used on Chandrayaan or drones, etc.
Q. How does Omnipresent use Machine vision, AI, ML, etc. in combination with drones? What kind of computational power and specifications does your drone have?
You can think of robots as a device or a machine that is gathering a lot of data. For example, in video analytics, the video can be taken from a camera, robot, or drone. Once the video comes, you have to make sense out of it. That's where the AI will be used. We've been deployed in a lot of factories, oil refineries, power plants, and at these places, we've built something a Nerve Center.
The nerve center takes in data from various platforms, whether it's a camera, drone, or robot. Then it processes those to generate the AI behavior which could be something like detecting whether somebody is wearing a mask or are people following social distancing, or in the factory, are people wearing safety gear. Similarly, there can be a variety of behaviors like detecting crowd behavior, fire, or anything that you think will require some kind of a noteworthy event that can be trained in the nerve center. It can then generate a notification that can go to the security head on his mobile phone. The processing can be done, whether the data is coming from a drone, a CCTV camera, robot, or even a mobile phone. Sometimes the video taken from the mobile phone can also be uploaded to our nerve center. Basically, the nerve center brings about data from all these sources, and then depending on the user's requirement, we add a certain type of behavior for them.
Q. Can you tell us how a typical solution like this is deployed? How do you train your AI and how do you deploy your solutions?
We can divide it into two parts. Sometimes clients want only the drone AI, sometimes they want the camera, and sometimes they want both. If they want to deploy drone-based AI, then we usually send a team of pilots and a drone. Right now, there are not enough drone manufacturers in India. So, we've been building our drones. Although, we think that we will see a lot of growth in software as opposed to hardware. We have focused quite a bit on the software part but we deploy a drone pilot at the site. We give a drone nerve center dashboard interface to the client. The client can request on the dashboard as to when they want to have a drone flight, in what region, and what are the kinds of things they want to see and then our pilot gets that message and they schedule a flight.
Once the drone collects the data, that data is then processed, usually on the cloud but sometimes, the client might have certain reservations, and they may not want the data to go outside the system. In that case, we can deploy a local server as well. So, the data gets processed on the local server and a report is generated and that report can again be viewed online by the client.
As we start getting more and more data, we also keep on training our engine. On day one, we have a pre-trained engine over there. But after a period of one or two months, once you've got enough data, we can also add more customized behaviors, which is something the clients might like. For example, in an oil refinery, it's very difficult to tell if there is a problem on a vertical structure like an oil flare. To inspect that, they may have to send somebody climbing up there, and they may even have to turn off the flare which causes a lot of losses to the client because they need to keep on running continuously. We do these kinds of inspections, and minute things like cracks, corrosion, structural damage, etc. can be detected by the AI and put on the report. One of the good places where the robots have a lot of good justifiable demands is dull, dirty, and dangerous tasks. For example, if you want to inspect a cell phone tower or a power line tower, or coal mines. This is where the machine should ideally go. That's when a machine is fully justified.
Q. Can you tell us about the Chandrayaan Project, the AI navigation software that was built by your company?
In Chandrayaan II, there was Pragyaan Rover but unfortunately, that rover didn't quite come out. But that rover will again be sent on Chandrayaan III. Once that rover lands on the Moon, it will take almost 10 minutes to upload one image on Moon or download one image from Moon, it's not going to be a 100 Gbps home connection. You have to do everything autonomously there.
The simulator software that we built for the Chandrayaan rover is the same software that the rover was supposed to use. Once the rover lands on the Moon, we see a simulated image of the rover. We can see a rover going on the Moon surface and we can also plan the path. We could click on any point and tell if the rover can cross a particular path or big rock. Also, we can know the distance, angle, height, etc. As the rover moves, we get the feedback and we can drive the rover sitting here. The rover sends us just two images and we make a single 3D image out of it. This is the software that the Chandrayaan rover is going to use, and once it lands on Moon, then our software is going to help it navigate and guide across the Moon.
Q. Do we get only two-dimensional data and how do you convert it into 3D? Do you have a LIDAR or Radar system also?
We don't have a LIDAR on Chandrayaan. It is very similar to how humans see 3D. The rover has two cameras like two eyes from which it sends the images. Initially, we wait for 10 minutes before the rover starts, then we download the two images sent by the rover. Once, we have converted the images in 3D, we get a clear picture. We can control the rover from Earth and know what it is doing on the Moon.
Q. As a company in India working on robotics and drones, what kind of technical challenges do you face? Do we have proper infrastructure for drone and robotics-based solutions?
One of the reasons why we are doing both hardware and software is because there is no standardization in this industry. As far as robots and drones are concerned, it's not like I can just buy one Intel-based computer and just load windows and then run my apps or whatever. As there's that lack of standardization, we probably end up doing more than what we need to do. One of the problems is to make sure that our hardware and software are compatible, and also generic enough. That presents a challenge. There is also a challenge in terms of hiring, we want to hire but we can't get the right kind of people. Maybe, jobs are slowing down in the software industry but in terms of robotics and AI, I think jobs are growing exponentially. We often end up training all our people, so that after a few months they start delivering.
Q. Omnipresent is one among the few companies to have got Govt. Approval to Start Ecommerce Drone Delivery trails. How are the trials going so far? When do you think E-commerce Drone Delivery will gain momentum?
We have done a drone delivery trial for one of our e-commerce partners, Shop X. We've also been working with a few other major e-commerce players. We did the drone delivery trial for a box of shoes being delivered from one warehouse to another warehouse in Bangalore. These two warehouses were about a few kilometers away. The government has selected a few companies and has granted permission for carrying drone delivery trials. As part of these trials, we will be attempting deliveries as large as 100 kilometers carrying a 15 kg payload. There are a lot of repercussions because the government has allowed drones up to 500kgs to be flown and they can be used in green zones. The green zones will practically cover more than 80% of the country and even in other zones, with special permissions; you will be able to fly. So with that kind of payload capacity, the drones will give much better costing and performance when compared to let's say, a small truck or a small vehicle carrying that kind of payload. These new regulations will cause a lot of disruption in the drone industry and a lot of new companies will grow and I think it's a very good time to get into this segment.
Q. What are the hurdles you're facing and why is drone delivery not a reality yet?
The biggest hurdles are regulations, which the government has just come out with. The new regulations seem very promising. It is still in draft, it will be ratified in about 30 - 45 days. Once that happens, we'll see a lot of investor interest in this segment; investors will love to get into this field, companies like Flipkart, Swiggy, Amazon, etc. will all love to get in here. Once these big names also start getting in them, you will automatically see this sort of snowballing effect, and we will see it becoming a reality very soon. Having said that, it will probably still not be like, a drone knocking on your window because there will still be a few security concerns, and the drones might have a designated landing from where a local boy/guard could deliver.
Q. How do you see the market for Video analytics solutions through Drones and Robots? Are companies and industries ready to adopt this new technology?
Already there are big companies like Tata, Adani, and Reliance that have got into drones. Big names are getting in here because they can sense the market and drone delivery is a huge market. After all, there's enough market to set up. A couple of more big delivery companies like DHL, a company FedEx can be set up just for drone deliveries. Defense is going to be also extremely big market because most of our future comebacks and things will start happening via robots and drones. The next big market is consumer drones. Currently, most consumer drones are coming from China and now with the government putting harder restrictions on Chinese drones, there's a big gap created here as to who is going to build all these drones for consumers in India. Drone-based agriculture will be another big sector because half of our landmass is agri-land and drones have been shown to boost agri yield by up to 20 - 25%. Imagine the kind of impact it can have on our GDP. We have around six lakh villages, and the drone can be operated locally. That means it can also generate employment, technicians, and pilots, and then they could serve the village using it for agriculture and also using it for other things like safety, security, and so forth. Just agriculture itself is going to be a huge segment.
The next big segment will be asset management for industries and construction. It's also useful for insurance companies. To assess how the structure was made, what is the progress, and then you know, maintenance of things like cell phone tower, electrical tower, pipelines, railway tracks, railroads. The National Highway Authority has just issued a circular saying that all our highway projects have to be now mandatorily be inspected and monitored via drones. There are a lot of segments that are going to get big; you will see 2-3 unicorns in drone space in India in the coming three to four years. I think it's a really good time to get in here.