Alastair has over 10 years’ experience leading research activities in scaled, high-growth industrial and technology markets. At Interact Analysis he is responsible for electric trucks and buses, autonomous trucks and off-highway electrification.
The rapid increase in use of sensors in automotive for safety applications and autonomous driving is providing the catalyst for much needed advances in the use of ‘sense and avoid’ for UAVs.
As millions of cars and trucks are fitted with advanced driver assistance systems (ADAS), the cost of the necessary sensors and processors that enable these functions is falling. In addition, the engineering and software know-how that is required for sensor fusion – the blending and interpretation of signals from different sensor types – is rapidly increasing.
Seeing what’s out there
Several UAV manufacturers already use ‘sense and avoid’ technologies. The albris drone from senseFly relies upon visual imagery and ultrasonic sensors to avoid obstacles; the Typhoon H from Yuneec employs Intel’s vision based RealSense Technology; and the Draganfly Commander uses a LiDAR rangefinder from LeddarTech.
In addition to ultrasound, visual and LiDAR technologies, there are RADAR solutions from companies like Aerotenna.
Challenges to be avoided
‘Sense and avoid’ is critical for the rapidly growing commercial UAV market. Fully autonomous flight operations will likely only be approved if UAV manufacturers can demonstrate safe, reliable ‘sense and avoid’. Furthermore, operators of UAVs will likely demand it as a means of protecting their investment and increasing the value of their UAV – autonomous flight lowers overall operational costs.
Whilst the technology is demonstrably useful, the challenge of using it more widely is three-fold: price, size and integration.
The cost of sensors – particularly LiDAR – is high. Whilst these may be applicable in very high-end UAVs, it may not be economically feasible to the lower end of the market to add these features.
The physical size and weight of such sensors also poses a problem. UAVs – particularly battery powered ones – have relatively short ranges and small payload capacities. The addition of one or more sensors will further limit operational usefulness despite the extra functionality that ‘sense and avoid’ brings.
Integration is possibly the biggest challenge though. To enable truly effective ‘sense and avoid’, multiple sensor types must be used – visual, RADAR and LiDAR. This will require sophisticated sensor fusion technology.
Clearing the way
The above challenges are not insurmountable; however, to achieve all three will take time and significant investment. Fortunately, these challenges are being overcome in the automotive industry on a daily basis.
With the uptake of ADAS technology in millions of vehicles, and many global manufacturers pursuing autonomous cars, the price and size of sensors are falling dramatically. For example, Velodyne, Quanergy and Innoviz – manufacturers of small, solid-state LiDAR systems are targeting sub-$100 sensors when produced in volume for automotive. Furthermore, the form factor of these sensors is much smaller than scanning LiDAR sensors. The UAV industry potentially stands to benefit as a result.
The auto industry also has many years of experience with sensor fusion. Many automotive tier 1s and semiconductor companies offer standard, off-the shelf sensor fusion products. Again, the volume in automotive is driving down the price of this technology to the benefit of the commercial UAV market.