The technology works in the event of a potential emergency and aims to be a safety measure that can contribute to accident prevention
IBM scientists have been granted a patent around a machine learning system that can dynamically shift control of an autonomous vehicle between a human driver and a vehicle control processor in the event of a potential emergency.
It claims the patent, Controlling driving modes of self-driving vehicles, will provide a safety measure that can contribute to the prevention of accidents. IBM researchers developed the patented system using their understanding of biological cognition and behaviour generation in the brain.
Their background as computational neuroscientists led the inventors to devise a cognitive model and technique that employs sensors and artificial intelligence to dynamically determine potential safety concerns and control whether self-driving vehicles are operated autonomously or by relinquishing control to a human driver.
For example, if a self-driving vehicle experiences an operational anomaly -- such as a faulty braking system, a burned out headlight, poor visibility and/or road conditions -- a comparison may be made by the system as to whether the on-board self-driving vehicle control processor or a human driver is in a better position to handle the operational anomaly.
If the comparison determines that the vehicle control processor is better able to handle the anomaly, the vehicle is placed in autonomous mode.
"Self-driving vehicles hold great promise and potential, but protecting the safety of passengers and other drivers remains a top priority for vehicle developers and manufacturers," said James Kozloski, manager, computational neuroscience and multiscale brain modelling, IBM Research and co-inventor on the patent.
He added: "We are focused on finding new ways to leverage our understanding of the human brain and inventing systems that can help those enterprises improve the safety of autonomous vehicles on the road."
While IBM’s newly patented machine learning invention addresses the complexity of dynamically enabling safe operation modes of an autonomous vehicle, other patented IBM inventions are focused on helping self-driving vehicles better anticipate and respond to actions of human drivers.
For example, Automatic driver modelling for integration of human-controlled vehicles into an autonomous vehicle network describes a machine learning system that models human driving techniques.
The invention also employs a common interface that enables self-driving vehicles to communicate with one another, learn and better understand how to interact with human drivers as the models become more informed.
According to the IBM Institute for Business Value, automobiles are evolving from a mode of transport to a moving data centre, outfitted with sensors and computers that capture information about the vehicle, its driver, occupants and surroundings.
At the same time, conversational interfaces are enabling drivers to interact with their vehicles more naturally and, with machine learning, automobiles can learn about their drivers and personalise the driving experience accordingly.
IBM inventors have patented numerous inventions that, among other things, can help vehicles become:
IBM has topped the list of US patent recipients for 24 consecutive years.
If you like this, you might be interested in reading the following:
BMW places IBM Watson in the passenger seat
Companies will work together to explore the role of Watson cognitive computing in personalising drivers’ experience
Ford drives towards autonomous leadership
The company is investing $1 billion during the next five years in Argo AI to combine its expertise with the speed of a start-up to further advance autonomous vehicles
Nevada’s autonomous vision
Nevada Department of Transportation is working with CH2M to develop a policy framework for driverless and connected vehicle initiatives in the state