AI is being developed quicker than ever as we gradually start letting go of our steering wheels, one white-knuckled finger at a time. However, it may be a while before we see full automation in our everyday vehicles.
The AI automotive market was valued at $783m in 2017 and is expected to reach close to $1.5bn by 2025, when the installation rate of AI-based systems of new vehicles will have risen by 109%.
Autonomous cars need to be able to identify potential hazards. The first and most important element of driver-free-driving is the car’s ability to anticipate the actions of others and make decisions of their own. Machine Learning is the beginning of this process, but we need to have developed full neural networks in vehicles’ CPUs before the practice is anywhere near safe.
Machine Learning relies on information from the real world and for the most part, that means sensors. Most of the sensor technology is actually already here.
We use radar for long range use, you may already have this technology installed in the car you drive today, like adaptive cruise control. But developers are aiming for up to 400 metres in range. Lidar and stereo video cameras will be used in conjunction with radar for mid-range detection. Ultrasonic sensors and short-range cameras are solutions that will be implemented for close-range situations. Until a solution as to how vehicles ‘see’ the world is fine-tuned, no automotive manufacturer will risk putting fully autonomous vehicles on the road.
Autonomous vehicles must also be able to pinpoint their location down to the nth degree. This means that sensor data must be used to model highly accurate 3D maps.
IoT and Data
IoT is currently being used within the automotive industry. Onboard CPUs are doubling up as IoT gateways, meaning an enormous quantity of data is being generated, collected and sent to the cloud for processing. Data is being captured on:
- A vehicle’s route
- The vehicle’s onboard diagnostics
In the not too distant future, all cars will be IoT enabled and real-time data will feed the ever-growing intelligence of the autonomous vehicle AI and traffic and incident data will be shared between connected cars.
According to Gartner, the automotive IoT market will grow to 5.8 billion endpoints in 2020, the growth will be owed, in part, to the increase in vehicles coming out of production with embedded IoT connectivity as standard. This will be supplemented by a range of add-on devices to accomplish specific tasks, such as fleet management in the commercial sphere. Smart infrastructure is also being developed in tandem with autonomous vehicle technology to enable V2X (vehicle-to-everything communication) systems.
The role of AI in Automotive Manufacturing
AI has applications that span the breadth of the automotive manufacturing sector. Manufacturers can implement Machine Learning into the very first step of auto-development: design. Data derived from car-users can give insight into the very design of a new line of cars or trucks. In the fabrication and assembly lines, AI-driven systems can be used to create schedules and manage workflows as well as monitor and control the efficiency of the operation itself.
Manufacturers can also reduce costs and downtime in production lines through Predictive Maintenance schedules with insight derived from the real-time monitoring of the factory as a whole.
QC and compliance are huge concerns of any automotive manufacturer and AI also lends a hand in that field. The constant feed of data ensures that systems get better over time at identifying defects. AI-powered hardware is at the point where automated visual inspections can be carried out to identify defects on components and provide superior QC and compliance to manufacturing regulations. It’s a really interesting time for the automotive industry as the latest technology bleeds into its every vein, from design right through to connectivity. We’re excited to play a role in the industry and see what the next developments in technology will bring.