Thought Leaders

The Future of Drone Technology - Autonomy, Collision Avoidance and Advanced Sensors

Drone technology has developed rapidly since the beginning of the century; driven by advances in sensing, electronics and engineering. Drones are expected to have an impact on every aspect of our lives - from delivering our shopping to collecting data for important scientific research. However, there are still several obstacles to overcome before this becomes a reality - how will drones accurately sense their environment and how will they be mass produced?

AZoSensors spoke to Rai Gohalwar, a UAV developer and enthusiast, about how these questions can be answered, the future of drone technology in general, and how mass collaboration may be what takes drone technology to the next level.  

What obstacles are preventing the widespread uptake of drones?

Drones are one of the fastest growing areas of technology in the world. However, as history has shown us many times, mass adoption of a product or an idea tends to be held back by government regulations.

Let’s look at cars for example. There was a time where there were lots of cars on the streets yet no laws regulating them. People would get into accidents often and it was virtually impossible to keep track of them. The government slowly caught up to the booming auto industry and regulated the market by creating rules around their operation and production.

Similarly, drones are waiting for the regulatory forces around the world to catch up. Currently, there are no manufacturing guidelines or regulations and every country has their own rules about flying. Standardization of the drone construction process would allow more collaborative efforts to develop the technology and help maintain a certain standard amongst all drones. Currently work on drone technology and their quality is very much every man (or company) for themselves.

The other factor holding back the widespread uptake of drones is competition. This is one of the only technology industries where one company (DJI) owns close to 80% of all sales. I respect DJI for their constant innovation and their ability to build well-crafted drones; however, monopolies are never good for the customer.

If another big drone manufacturer existed there would be more of a drive to innovate and offer competitive prices, both companies would be forced to innovate and would have to offer competitive prices. For example, if Indian Motors were not around during the early 1900’s, Harley Davidson would not be as big as they are today. The fact that they had to “beat” Indian in the motorcycle race pushed them toward clever innovation.

The drone industry lacks the competition required for true innovation. Shutterstock | Pressmaster

How crucial is the improvement of obstacle avoidance for autonomous drones? How do you expect this technology to be improved?

Obstacle avoidance is a technology that will propel drone autonomy forward. In the near future we will increasingly rely on drones for industrial and commercial tasks. Drones will need to operate in challenging environments without any human interaction, and the machine’s ability to avoid obstacles and manoeuvre around them will be necessary.

Deliveries, mine inspection, construction, and more, are all fields where drones need to be aware of their environment. On a consumer level, obstacle avoidance helps keep the drones from crashing and makes them “idiot proof”, in a similar way to automatic front and rear braking in modern cars. These features prevent the driver from crashing either if they are not paying attention to the road or if they’re not a skilled driver. Similarly, obstacle avoidance in drones prevents the pilot from crashing if they lose sight, orientation or are unfamiliar with the controls.

Obstacle avoidance systems are rapidly developing. There is a lot of research in academia on visual avoidance systems, some using pre-existing technologies such as the X-Box Kinect. Industry is also working on other solutions.

Companies such as DJI have incorporated several front, down, top and side facing sensors, which all communicate with each other to detect the drones surroundings. Whilst collision avoidance has certainly got better over the past few years there is still a lot of room for improvement. The problem with most obstacle avoidance systems today is that there is only one data entity per direction. This is ok to detect walls and big objects, but it’s unreliable when trying to manoeuvre around small objects such as a pole.

Multi-segment LiDAR, or a spinning single beam LiDAR is going to be the technology that changes the drone obstacle avoidance landscape. LiDAR allows drones to both detect their surroundings and to figure out a way to go around them.

Which other drone components must be improved before autonomous drones can become a reality?

There are a few other key components required for a truly autonomous drone.

The first is GPS. The standard GPS that is currently used on most drones is accurate to a distance of around two meters, and this changes drastically depending on the environment. For autonomous capabilities, drones will need to use more accurate navigation systems such as RTK GPS.

Lots of drones have optical flow sensors underneath them, which helps with stable flight at low altitudes (up to 3 meters). An accurate navigation system, such as RTK GPS, ensures that autonomous drones takes their intended path. This is especially important if you consider a scenario where there are hundreds of drones flying around. There are an increasing number of manufacturers that now produce RTK GPS for drones. Swift Navigation and Emlid both have fantastic RTK GNSS units that work well on drones.

Another piece of technology that I feel is important but doesn’t get the attention it needs is the battery technology. As we become more reliant on drones, we will need them to go further and to stay in the air for longer.

Currently, most drones use lithium-polymer cells because of their good energy density and high discharge rate. Though they are harder to maintain compared to lithium-ion and other batteries, advanced BMS (battery management systems) technology has made them very easy to use. I’ve been looking at alternatives for a while and the most promising is lithium-sulfur cells. They have a much greater energy density than lithium-polymer and are much safer.

Recently scientists have improved this technology and within the next few years, these cells will be replacing lithium-ion and polymer cells.

To travel the distances required for applications such as surveying advances in battery technology are required. Shutterstock | ZakiraZ

What features of LiDAR sensors make them the best option for obstacle avoidance?  Are there any unique features of the LeddarOne that make it superior to other LiDAR sensors?

LiDAR is an amazing piece of technology. This technology has come a long way; we now have tiny LiDAR sensors which can demonstrate long ranges at a low power consumption.

Most people don't realise this, but the strongest and most unique aspect of LiDAR is the field of view (FOV) of the sensor. With an extremely narrow FOV, which is typically 3 degrees, the sensor is capable of detecting small objects from very far away.  LiDAR is great for obstacle avoidance because of its range (typically 40m) and reliability. LiDAR sensors can also work in many harsh environments and function in low light.

LeddarOne is a unique sensor. I spent some time tinkering with it and also deployed it on my own drone. What stood out to me is the attention to detail used in the sensor; LeddarTech had thought of everything.

There are some distinct features of this sensor that make it apart from other similar sensors out there. Firstly, the multiple output array of diodes helps to significantly increase the robustness of the sensor. This is in comparison to most other LiDAR sensors which have only one out beam. With the LeddarOne I covered all but one of the output diodes and the sensor still worked flawlessly.

The other strength of the LeddarOne is that it communicates using a serial TTL bus. This ensures accurate data which is also relative. Similar LiDAR sensors out use analog connectivity, which can introduce discrepancies in the data and introduce offsets which means the sensors need to be calibrated often to remain error free. By using a digital method of communication the LeddarOne solves this problem.

What future applications do you see autonomous drones being used in?

The beauty of drone technology is that the applications are endless. Everyday I see a new drone application that someone is working on.

Deliveries using drones are going to be one of the biggest drivers of drone technology. Whether its groceries, medicine, cooked food, consumer products, or even people; drones are going to be used to move things from one place to another quickly, safely and with minimal effort. Amazon, Google and other big companies have already started deploying the necessary infrastructure for this.

That said, there are still some large obstacles to be overcome. For example, before we fill the sky with drones flying in every conceivable direction a universal communication protocol, as in aeroplanes, needs to be developed.  

Drones are already being heavily used in agriculture, construction, and other surveying fields for mapping. Advances in drone technology have made collecting data from the sky far easier and cheaper. I would not be surprised if drones become the industry standard for mapping and precision agriculture.

Drone technology is already being used for agricultural surveying. Shutterstock | Budimir Jevtic​

What was it about drone technology that initially made it of interest to you? How did you first begin to develop that interest?

I’ve was lucky to have found my passion at a young age. I used to love playing around with motors as a child and this led into me getting into RC airplanes. I grew up with the ‘.com’ explosion so I could access a lot of information that would have been hard to find 10 years earlier. I distinctly remember watching Dave from RCPowers on YouTube. He was a pioneer of the RC community online. I learnt from his videos and started designing my own airplanes, soon I could fly my own designs.

Around high school graduation I learned about multi-copters. I first built a tricopter out of balsa which was using an old Hobbyking KK board. It crashed minutes after takeoff much like my other experiments. I kept scraping and building and got good at it. Shortly after I got a job working at the agriculture research department in Ottawa. There I was exposed to the industrial applications of drones. I never knew they could be used to collect data. I was only flying around for fun.

It was here that I was approached by many scientists working there who had interesting ideas of drone applications. Most of my time was spent on building a drone to collect NDVI and thermal images of soybean plants. That was also the time I got serious about my online presence. I started uploading videos on YouTube about drones.

How do you expect user experimentation, open source programming and non-commercial collaborations to play a role in the evolution of drone technology?

I consider Open-Source to be the revolution of the modern generation. For the first time ever, code hosting sites such as Github.com have allowed people around the world to be exposed to collaborative technology projects.

One of the biggest open source drone platforms is ArduPilot. As the name hints, it was originally a arduino project which evolved into its own thing. There are hundreds of people who contribute and learn from the ardupilot project. Open source so powerful because the people who commit to projects are usually the ones most passionate about it This fuels creativity and new innovation.

A great example of this would be Andrew Tridgell. He was introduced to ArduPilot a few years ago and has completely re-defined the project. He created a HAL (Hardware Abstraction Layer) over ArduPilot which gave way to so many other people to develop their own flight controller boards.

I should conclude by saying that open source projects such as ArduPilot and Paparazzi allow the world to mold the software according to its needs. If a company develops a closed source flight controller, it will only consist of features that a select bunch of people deemed necessary. Open source usually runs a few months ahead of big companies. People create and commit features that they want to see on drones.

Hobbyists and professionals collaborating together continue to have a strong impact on the future direction of drone technology. Shutterstock | aerogondo2

From a drone user’s perspective – if you could give one piece of advice to manufacturers what would it be?

I consider myself a novice when it comes to consumer drones. I have experience building my own drones but I am not a big advocate of buying drones. In fact, I only own 1 drone that I didn’t build myself, a DJI Mavic Pro.

The drone industry is a strange one. There are always these small companies trying to build a drone which ends up getting swallowed up by the larger companies in the business. One key thing that most manufacturers forget to address is attention to detail. Customers subconsciously looks for a good design before functionality.

Most people don’t want to modify their products. They just want something that is a good design, and does what it should without asking too many questions. Most drone manufacturers forget to build a drone around a user. They build a drone around an application and force the user to go through a steep learning curve. I think a good drone should allow the user to do what they want it to without the user having to learn anything about drones. When you buy an iPod, you don’t read the instruction manual. You just turn it on and use it.

On a technical note, I would recommend all drone manufacturers to use an optical flow sensor and forward proximity sensor on their drones. These 2 sensors can drastically improve a drones performance. My last piece of advice to manufacturers would be “have fun designing”. It seems like all the commercial drones look the same. Not many manufacturers experiment with new designs.

How can sensor providers like LeddarTech help in the advancement of drone technology?

Sensors are at the heart of all drones. Companies like LeddarTech are in a unique position. They are the fuel for autonomy and innovation in drones. Their ability to see the future of a technology and build tools around that today is so necessary.

Consider drone deliveries; sensor companies need to see 10 years ahead and envision the types of drones used then. What kind of sensors will they need? What kind of power and weight specifications will the industry require? What will the drone be used for? Will the sensor be good enough to replace currently used solutions?

At its core, a drone (miniature UAV) is best suited as a data collection machine. Creating the sensors to collect data is in the hands of companies like LeddarTech. Not only do they have to invent something small and lightweight, it must also meet industry standards as far as the quality of data goes. One mistake I often see sensor companies make is their lack of flexibility. I love LeddarTech because they are very integrated with the community the cater to. They listen to their consumers and their products are designed around the consumer. They are not afraid to push the norm.

Where can our readers find out more about your thoughts on drone technology and the LeddarOne LiDAR sensor?

I have a YouTube channel where I upload some of my projects. I recently reviewed the LeddarOne which you can view here. Most of my videos are just experiments I try with drones.

LeddarTech also run an awesome blog which showcases their latest technology which you can find on their website.

About Rai Gohalwar

Rai Gohlawar is a drone developer and enthusiast who runs his own popular YouTube channel. Rai developed a love of drones from an early age, entering a field as a hobbyist who constructed his own systems. This passion led Rai to study engineering at the University of Waterloo.

Following his time at University Rai worked at AgriCan Canada, an agricultural research centre, where he developed novel drone based solutions for crop monitoring and agricultural analysis. In addition to this Rai has also worked as a UAV engineer for Swift Navigations, before continuing his studies in Mechatronics Engineering at Simon Fraser University.

In his spare time Rai continues to run his YouTube channel whilst also working on his own drone projects.

Disclaimer: The views expressed here are those of the interviewee and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.

Ask A Question

Do you have a question you'd like to ask regarding this article?

Leave your feedback
Submit