How AI is driving the automotive industry

[yosoytono]

The Electronic Highway, an IEEE Spectrum article written in the early 1970s by engineers Robert Fenton and Karl Olson, hypothesized that the future of automated vehicles would be dependent on a smart infrastructure that would "drive" cars on roadways. However, the lack of technical resources at that time kept them away from becoming their idea into reality.  

The past decade brought an explosion of vehicle technology within the automotive industry; most of the factors leading to AI shaped up, nowadays Artificial Intelligence is essentially transforming our reality. Car makers are using AI and its technologies to develop better-improved products & solutions for their customer base.  

The automotive industry is empowering vehicles to act like a human driver and respond to different situations: vehicles with AI are gathering, processing, and recognizing information from their environment using sensory inputs. Based on the collected data, vehicles are evaluating and learning the contextual implications of choosing a specific action when facing a hurdle. 

While AI automotive applications are normally related to autonomous cars —also known as driverless cars, self-driving cars, or robotic cars; is only one of many uses for Artificial Intelligence in the automotive industry. Car makers want vehicles not only to include smart features like self-driving programs, but also built-in smart assistants & smart amenities to ensure driver accessibility and comfort, or even impact detection & damage prevention. 

Nowadays, vehicles are outfitted with advanced tools (long-range radar, LIDAR, cameras, short/medium-range radar, ultrasound...) and also applying computer technology and software engineering to build a three-dimensional map of all the activities that happen around the vehicle. Basically, AI is driving the automotive industry by using these technologies:

1. Machine Learning (ML):

Machine Learning is actually a subset of Artificial Intelligence: devices with AI would do tasks in a way humans consider smart. ML is an application of AI where machines are given certain data and they learn for themselves. 

Toyota has gone further combining Big Data (BD), ML, and AI to create highly responsive autonomous systems that help in mobility for less-able drivers. 

2. Deep Learning (DL):

Deep Learning is the process where ML is implemented: DL breaks down tasks in manageable chunks and helps AI activities to happen without setbacks. DL allows the vehicle to learn and mimic the activities in the neuron layers of the brain. 

DL is very useful for Advanced Driving Assistance Systems (ADAS) and Autonomous Driving (AD).

3. Cognitive Capabilities (CC):

Cognitive Analysis imitates the human behavior by crossing behavior patterns with (unstructured) data mining to get insights. Cognitive Systems aims to perform just like a human would react in a real-life situation.

BMW has partnered with IBM to add cognitive capabilities to its cars: they are using Big Blue’s Watson AI technology to help vehicles communicate with each other.

4. Internet of Things (IoT):

Newly-manufactured vehicles have many embedded devices, smart sensors, geo-analytic capabilities, BD enabled systems, and connectivity applications that would help vehicle owners to have: firmware updates and performance-related issues correction through over-the-air software, performance data reports sent to the manufacturer/dealer/service center, safety parameter management, enhanced manufacturing quality, and quicker smart responses in case of emergencies.

Car makers have noticed a growing trend and a significant business opportunity for connecting their vehicles. It is expected +380 million connected cars by 2020, up from 36 million in 2015. 

5. Infotainment Systems:

Today, infotainment human-machine interfaces are already interacting with AI features to provide in-vehicle solutions: speech and gesture recognition, eye tracking, monitoring driving, and natural-language database management.

The demand for high-quality hardware and software has been rising in the past years to integrate more complex AI solutions that can interact with algorithms collected from cloud-based neural networks: driver condition evaluation, Vehicle-to-Everything (V2X) communication, sensor fusion engine control, camera-attached machine vision systems, and radar-based detection.

Automotive AI is moving to a new level of excellence within the automotive industry by using the huge computational power available today. The "autonomous cars" phenomenon is not just about self-driving vehicles and how they act as a real driver, but on how AI is helping to build cars that can "feel" their environment and navigate through obstacles that come up during driving.