Where is AI in the auto industry now – and where is it heading?
The shape-shifting, artificial intelligence (AI)-powered cars seen in futuristic films aren't appearing anytime soon but AI technology certainly exists today in the auto market and will be increasingly visible in the future.
Industry observers are saying that AI and machine learning have reached a tipping point and will cause enormous advances through the next few years in many markets. In the automotive sector, these technologies will be transformative. Analysis firm IHS Markit predicted that the installation rate of AI-based systems in new vehicles would rise by 109% in 2025, compared to a modest 8% adoption rate in 2015. In February 2017, Ford paid $1 billion for artificial intelligence start-up Argo AI.
Over the next five years alone, AI-based systems will become standard in the areas of infotainment systems including speech and gesture recognition, eye tracking and driver monitoring, along with Advanced Driver Assistance Systems (ADAS) and autonomous vehicles using camera-based machine vision, radar-based direction units, driver condition evaluation and sensor fusion engine control units.
In tracking the future role of AI in the auto industry, it's useful to focus on where these predictive, smart technologies already exist and how they will expand their functionality.
With the ability of computer systems to improve their performance by exposure to data without the need to follow explicitly programmed instructions, machine learning is already becoming so pervasive that many of us probably use it every day without knowing it. This technology is a key component in the driverless cars now cruising down some roadways in pilot projects as well as in actual autonomous vehicles and related services from companies like Uber and Volvo.
One ripe area for machine learning in vehicles is cybersecurity systems, which have self-learning and self-healing capabilities. Some experts say we could start to see this technology appearing in enterprises in 2017-18. The ultimate aim for automakers are AI-based security systems that are self adapting and self defending with ways to guard against hacks and new threats without any humans needing to program the system.
Already common in everything from our smartphones to those annoying call centers, voice recognition is in our vehicles now and this trend will continue unabated - most recently seen in Ford adopting Amazon’s Alexa for Sync, a voice-controlled communications and entertainment system. However, there are challenges in moving from today's level of speech recognition to technology that can truly understand context and thus serve a driver's needs safely and accurately.
Companies are busily working on this problem, adding intelligence to AI systems so that cars will be able to more deeply understand context. Technologies such as extensive parallel processing, advanced algorithms and massive data sets to feed the algorithms in future cars will enable vehicles to continue learning as we drive them.
Computer vision - the ability to identify objects, scenes and activities in unconstrained environments - is one of the key technologies in today's autonomous vehicles and will play a key role as well to support drivers of AI-equipped cars.
Computer vision in today's driverless cars from companies like Google demonstrates how this AI technology works, although the car's "eyes" - just like in humans - depend upon the "brains" found in massive compute power, complex algorithms and deep learning. The vision in such a car is handled by cameras to detect traffic lights, signs and moving objects like pedestrians and bicyclists. Meanwhile, radar sensors and lidar units send beams to measure the distance to obstacles.
Of the various AI technologies in vehicles, computer vision is the most complicated and advanced. Given the profound safety requirements of cars, whether manned or unmanned, this isn't surprising but various vision features are already out there. These include things like collision avoidance systems, drifting warnings, blind-spot detectors, enhanced cruise control and self-parking but the capabilities of computer vision in cars are just beginning.
Involving highly refined algorithms, databases and powerful computing capabilities, optimisation technology has been used in the business sector for years to predict and improve outcomes. Optimisation technology can also benefit vehicles by helping cars automate complex decisions and make trade-offs to best use limited resources like system bandwidth. An example is algorithms to help car systems make intelligent choices by selectively pulling only relevant data from the cloud to improve fast, accurate decisions.
Another example is using adaptive delta compression to improve a car's ability to keep its software updated. This near-future technique involves efficient differential compression algorithms that take a delta, or status, of current car software systems and adapt updates according to the needs of a particular car at a particular point in time. With incremental changes to software, this concept takes what is already there and adds the differential, achieving faster, smaller updates.
Another important technology long used in the commercial sector, business rules engines are programmed to follow defined orders to make smarter, faster decisions. In the vehicle world - and where AI is concerned - rule-based systems add a layer of intelligence. As cars receive more and more data, the rules adjust to become better and more granular over time.
Consider rules in adaptive cruise control, in which the car will take the data from its surroundings and make automated decisions but will have the ability to improve these decisions. Automated parking and adapted automated parking are examples.
Planning and scheduling technologies
Yet another key tool widely deployed in the business sector, planning and scheduling systems utilise optimisation and other algorithm-driven software to automatically define a sequence of events to meet defined goals.
Such systems are already used by car manufacturers for their supply chains but as the technology moves forward, they will appear in other areas such as improving over-the-air automatic software updates to a vehicle's many electronic control units. For example, planning and scheduling systems can determine the status of a car's software and define the best update sequence. This is an area ripe for AI technology.
Robotics on the manufacturing floor have transformed vehicle production. Robots have been used in auto manufacturing for years but the increasingly intelligent robots being produced today and tomorrow will have far-reaching impact on the industry. With their ever-improving cognitive abilities, smart robots will be introduced to undertake more complex problems and will contain intelligence that enables decision making. It’s predicted now that robotics will move from the manufacturing assembly line to the dealership with the goal being that the robots will assess, analyse and then repair vehicles module by module.
Industry analysts are unanimous about the increasing intelligence of cars in years to come. These smart, networked computers on wheels will improve our lives in many ways. Gartner Group predicts that in just four years, there will be a quarter of a billion connected vehicles on the road, which will make possible new in-vehicle services and expanded automated driving capabilities. AI technology will enable tomorrow's car to better understand our needs and be able to react to and learn from us just as fellow humans do. Ideally, AI-driven cars will improve on us humans and actually learn from the mistakes they make.
Imagine a car with windows that can show the outside view or display movies or reveal information, depending on our wishes. We may be able to rotate our seats however we desire, including making them into comfortable beds. In fact, future autonomous cars might have such advanced built-in safety features that the seatbelt might not even be needed. This brave new automotive world may move beyond being a science fiction movie staple into reality within most of our lifetimes.
- » Munich Re acquires IoT middleware platform provider Relayr for $300 million
- » Why the IoT needs more of an edge: Edge computing and the data centre
- » Cloudera teams up with Red Hat and Eurotech for end-to-end open source IoT architecture
- » Opinion: Television still has the power to connect the home
- » Samsung outlines vision for connected living through AI, IoT and 5G at IFA