Five years ago, IBM Watson triumphed over human competitors to win the TV game show Jeopardy! Subsequently, Watson has been busy in fields including healthcare, medicine, and genetics.
Now, researchers at Fuji Heavy Industries (FHI), the manufacturer of Subaru automobiles, and IBM Japan are collaborating to put Watson in the driver’s seat in the role of an assistant if not an autonomous operator.
By way of background, Forester, Impreza, Legacy, Outback, WRX, and Crosstrek Subaru models are available with the EyeSight driver-assist system, which uses stereo cameras mounted near the rearview mirror (top center of the photograph) to detect other vehicles, pedestrians, cyclists, and motorcyclists. In addition to avoiding frontal collisions and limiting damage in the event of an accident, EyeSight supports functions such as adaptive cruise control, active lane keep, and precollision braking and throttle management. Subaru first brought EyeSight to the United States with the 2013 model year Legacy.1
Courtesy of Subaru/Fuji Heavy Industries
At a press conference March 23 at the 2016 New York International Auto Show, Yasuyuki Yoshinaga, FHI president and CEO, introduced the new Impreza and described it as “the first of Subaru’s next-generation vehicles,” with EyeSight and other features constituting “the bedrock of our core technologies.”
The collaborative effort between FHI and IBM Japan is aimed at helping FHI engineers manage the vast amounts of test image data associated with EyeSight. A system put in place in April allows FHI engineers to search for and analyze test images to help speed the development of advanced driver-assist technologies.
In the works is a system that will leverage IBM Watson Internet of Things for Automotive to evaluate the applicability of cloud technology to advanced drive-assist technology. IBM IoT for Automotive supports real-time analytics by integrating a range of other data with data collected from sensors mounted on each car.
Apart from the initiative involving FHI, IBM Watson IoT for Automotive is finding use in several applications. Daimler subsidiary moovel GmbH has created a car-sharing app, and truck maker Scania has used the technology to improve the driver response time of fleets through in-vehicle social messaging.
In addition, Panasonic Automotive Systems, which makes head units, amplifiers, and radios, has employed IBM cloud technology to collaborate with global partners and customers. In a video posted on the IBM Watson IoT for Automotive website, John Penoyer, group manager, tools, says, “In a 24-hour turnaround cycle, we were able to close 200 defects. We’re linking together requirements management, systems engineering, and software development” in order to build an all-in-one very tight iterative process.
Looking toward Watson IoT applications beyond automotive, Avnet recently debuted the MicroZed IoT Starter Kit, which includes a Xilinx Zynq-7000 All Programmable SoC and pluggable sensors from Maxim Integrated and STMicroelectronics. The kit integrates the IBM Watson IoT Platform agent on top of a custom-configured, certified image of the Wind River Pulsar Linux operating system. Using a standard MQTT messaging protocol, IBM’s Watson IoT Platform agent enables registered, secure connection to the Watson IoT Platform and additional cloud services from the IBM Bluemix portfolio.
“Avnet’s MicroZed Industrial IoT Starter Kit, supported by IBM’s Watson IoT Platform, enables IoT developers throughout the channel ecosystem to create new products that offer a differentiated level of user interaction and experience,” said Jack Desjardins, vice president, Alliances & Ecosystem Business Development, IBM Watson IoT, in a press release. “By simplifying IoT device-to-cloud integration with this kit, IBM and Avnet are helping to inspire a new generation of cognitive IoT systems and applications that have the capability to learn from the physical world.”
Watson also has taken to a virtual classroom in the form of “Jill Watson,” a robotic teaching assistant at Georgia Tech that helps students taking an online artificial intelligence course.
As reported by Melissa Korn in The Wall Street Journal,2 Ashok Goel, a professor of computer science, says students in the online class can post 10,000 questions per semester, straining the capabilities of the human TAs. Researchers trained Ms. Watson using a database of 40,000 discussion forum posts; she responds to questions if she has a 97% confidence level that her answer will be correct.
One student was about to nominate Ms. Watson for an “outstanding TA” award just before learning of her true nature.
Korn quotes another satisfied student as saying, “We’re taking an artificial intelligence class. There should be some artificial intelligence here.”
References
- Stoklosa, A., “We Try Out Subaru’s EyeSight Collision-Avoidance Tech,” Car and Driver, Sept. 7, 2012.
- Korn, M., “Imagine Discovering That Your Teaching Assistant Really Is a Robot,” The Wall Street Journal, May 6, 2016.
