Intelligent driver-assistive technology

再生 再生 再生
*This is a scene from the use of a driving simulator.

Striving for zero traffic collision fatalities:
Intelligent driver assistance that aims to eliminate human error
in driving

Development Story

Honda is striving to attain its goal of realizing “zero traffic collision fatalities involving Honda motorcycles and automobiles globally by 2050.” That said, no matter how much we advance safety technologies for motorcycles and automobiles, this alone will not eliminate traffic collision fatalities completely. We need to work on making people’s driving behavior safe by eliminating human errors.

When fully automated driving becomes available, there will no longer be errors made by human drivers. However, as a company that values the joy and freedom of mobility, Honda believes that driving performed by humans will continue to be important and, therefore, we take on challenges to make driving safe by expanding human capability with our intelligent driver-assistive technologies. Our ultimate goal is to realize a society where everyone can enjoy mobility with total peace of mind.

Striving for zero traffic collision fatalities

Giving compassionate
consideration to each person

To realize a society with zero traffic collision fatalities, we must eliminate collisions caused by human error, in other words, mistakes drivers make in operating their vehicles. To this end, we studied the human brain to make clear the true factors of human error. Our research showed that there is no one-size-fits-all solution to human error in driving and we need to address it while giving compassionate consideration to the personal situation of each driver.

The reason Honda began research into brain activity
and human error

Mistakes that drivers make in operating their vehicle, or human error, include overlooking the dangers and/or incorrect operation of the vehicle due to a lack of driving experience, and other errors related to the declining functionality of elderly drivers and/or changes in drivers’ physical condition. If we can eliminate human error while driving, we can reduce the number of collisions.

While conducting research into ways to improve the dynamic performance of our vehicles, we looked at the relationship between factors that enable the driver to feel the advancement of vehicle performance and driving behavior. That is how we got into brain research. From there, we began studying the relationship between driving behavior and traffic collisions. In order to understand the underlying factors of various driving behavior, we began research into the relationship between human brain activity and human error, and evolved our research into the area of driver-assistive technologies that prevent human error.

Complicated relationships
among the driver’s gaze movement,
brain activity and human error

After living organisms came into existence on earth with eye functions, it took approximately 500 million years before brain development began. When we learned this fact, we became eager to understand the brain that processes information from the eyes at a higher level and apply our study results to ensure safe driving by people. This became a major motivator for us to move ahead with our brain research.

Using the fMRI*, we studied brain activity and eye movement of people who are operating a driving simulator. The results showed that high-risk drivers have low spatial recognition capability, and their inability to see or anticipate dangers becomes a major risk factor. We also found that people judge risks based on memory and knowledge of what they experienced in the past.

*fMRI: Functional magnetic resonance imaging. One of the non-invasive methods to survey brain activity using MRI equipment.

An MRI machine in which the test subject can operate a driving simulator
The difference in brain activity between a regular driver and a driver at minimum risk who rarely encounters dangerous situations.
Driving a car involves the activity of various parts of the brain. High-risk drivers show weaker brain activity in these areas.

Setting normative driving models

Lead the driver to drive
at minimum risk

Since different people experience different kinds of human error, we first set up a normative driving model that minimizes the risk of collisions and then studied a system that assists each driver by filling the gap between their driving and the normative driving model.

AI leads the driver to drive at minimum risk

Since there are individual differences in risk related to safe driving, sample driving data that can serve as a yardstick is necessary to assist all drivers. Therefore, we set up a risk-minimizing normative driving model based on the brain activity of drivers who drive safely.

The basic concept of "intelligent driver assistance" is to 1) take data on the driver's gaze movements (line of sight), driving operations and surrounding road traffic conditions, 2) compare them with the normative driving model, 3) estimate the driver's risk level, and 4) provide assistance to fill the gap between the normative driving model and their driving.

The AI system calculates the assistance the driver needs to match their driving to the normative driving model on a real-time basis. The normative driving model incorporates the risk-minimizing characteristics of the eyes, brain, limb use and movement, and the AI system assists the driver when there is a time delay or insufficient driver eye movements or driving operations.

Based on eye movements and driving operations, the risk level of the driver is estimated in terms of safe driving. Then, AI makes a comparison with the normative driving model, and provides assistance to minimize the risk.

Expanding capability of high-risk drivers
with intelligent driver assistance

Our intelligent driver assistance is not designed to correct the driving of high-risk drivers. The goal is to provide HMI* and operational assistance that enables drivers to feel as if their driving skill has improved. For example, the system enables the driver to spot pedestrians and other objects that were not visible to the driver, to start steering at the right time while turning the curve and to control the accelerator pedal at the right timing in consideration of vehicles merging into the same lane. We believe that this type of assistance will eliminate various levels of anxiety drivers feel about driving and prompt a positive desire to go out on the road.

*Human machine interface: Means or devices for humans and machines to exchange information.

The narrow field of vision, one of the typical characteristics of inexperienced drivers, is widened to that of drivers at minimum risk in order to assist driving operations, thereby reducing human error. We are striving to expand human driving capability which enables drivers to feel as if their driving skill has improved.

Assisting driver’s ability with HMI that utilizes human intuition

Making human driving safe

Honda has developed a driving simulator equipped with an AI system for intelligent driver assistance. Having a variety of people experience the simulator, we are currently accumulating evaluation and driving data to further our research. Moreover, we are having people experience the intelligent driver assistance using our test vehicle equipped with some of the functions.

The driving simulator that provides driver assistance using HMI utilizing human instinct

Using a driving simulator equipped with an AI system for intelligent driver assistance, we are conducting an experiment in which participants experience driver assistance. The system estimates the probability of human error based on the driver’s gaze movements, vehicle operation data and driving data. Based on the estimate, the system provides real-time assistance for the driver in accordance with the situation. The assistance provided through an intuitive HMI included “cognitive assist” such as showing a message on the Head-up Display or retracting the seatbelt as well as an “operational assist” such as assisting with steering operation and applying reactive force to the accelerator pedal.

Based on the results of driving simulator experiments with various participants, we are verifying and advancing our risk estimation and driving assistance methods. From here onward, we will verify our normative driving model by driving the actual vehicle on actual roads and making necessary updates to the normative model.

■ Estimating human errors:・The direction of the driver’s gaze and facial orientation are being monitored with the driver monitoring camera.
・The driver’s steering wheel operations, pedal operations and vehicle behavior are being detected by sensors.
・AI system compares them with the normative driving model.
■ Providing driver assistance with HMI that utilizes human intuition:・Alerts the driver with message on the Head-up Display
・Alerts the driver by retracting the seatbelt
・Assisting the driver with steering wheel/pedal operations

Experiencing Honda’s intelligent driver assistance
using a test vehicle equipped with some of the system’s functions

We have developed a test vehicle equipped with an AI system and a driver monitoring camera that monitors the direction of the driver’s gaze and facial orientation. This vehicle also features a special camera mounted on the top of the roof that monitors the surroundings of the vehicle.
Based on the information on the driver’s gaze movements, data on driving operations and vehicle behavior, the AI system estimates the risk level of the driving and provide driver assistance to bring the driving close to the normative driving model.

The HMI on this vehicle includes a LED indicator to alert the driver to look to the left and right, a seatbelt retractor and a three-dimensional audio system that enables the driver to feel auditory sensations that indicate the direction the danger is coming from. We are verifying and advancing the system by testing the driver assistance using the actual vehicle.

The test vehicle is equipped with a part of the intelligent driver-assistive system
One example of driver assistance. The presence of a pedestrian detected by the camera will be notified to the driver by the risk indicator, before the vehicle comes close to the pedestrian.

Passion to make driving safe for people all around the world

Our research into the relationship between driving and brain activity identified that drivers at high risk levels in terms of safe driving have a narrow field of vision and low spatial recognition capability. Also, we learned that human memory is another important factor in driving as people drive in a way that they believe will reduce risk in light of their memory and knowledge obtained from their past experiences.

Our intelligent driver assistance will enable drivers to recognize risks that they couldn’t see before and drive safely and smoothly, which gives them a sense of accomplishment and satisfaction. When anxiety about driving is eliminated and driving brings the joy to each person, we believe that people will want to keep experiencing the joy and freedom of mobility. Based on this belief, we will utilize Honda’s existing global network for driving safety promotion activities and work with people who drive in the safest manner in each country to develop normative driving models tailored to the traffic conditions of each respective country around the world. In this way, Honda wants to make driving safe and with peace mind for each and every customer around the world.

Minoru Higuchi
Innovative Research Excellence
Honda R&D Co., Ltd.

Driver-assistive technologies project leader